Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
#define ICustomStreamSink StreamSink
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#ifndef __cplusplus_winrt
|
|
|
|
|
|
|
|
|
|
#define __is_winrt_array(type) (type == ABI::Windows::Foundation::PropertyType::PropertyType_UInt8Array || type == ABI::Windows::Foundation::PropertyType::PropertyType_Int16Array ||\
|
|
|
|
|
type == ABI::Windows::Foundation::PropertyType::PropertyType_UInt16Array || type == ABI::Windows::Foundation::PropertyType::PropertyType_Int32Array ||\
|
|
|
|
|
type == ABI::Windows::Foundation::PropertyType::PropertyType_UInt32Array || type == ABI::Windows::Foundation::PropertyType::PropertyType_Int64Array ||\
|
|
|
|
|
type == ABI::Windows::Foundation::PropertyType::PropertyType_UInt64Array || type == ABI::Windows::Foundation::PropertyType::PropertyType_SingleArray ||\
|
|
|
|
|
type == ABI::Windows::Foundation::PropertyType::PropertyType_DoubleArray || type == ABI::Windows::Foundation::PropertyType::PropertyType_Char16Array ||\
|
|
|
|
|
type == ABI::Windows::Foundation::PropertyType::PropertyType_BooleanArray || type == ABI::Windows::Foundation::PropertyType::PropertyType_StringArray ||\
|
|
|
|
|
type == ABI::Windows::Foundation::PropertyType::PropertyType_InspectableArray || type == ABI::Windows::Foundation::PropertyType::PropertyType_DateTimeArray ||\
|
|
|
|
|
type == ABI::Windows::Foundation::PropertyType::PropertyType_TimeSpanArray || type == ABI::Windows::Foundation::PropertyType::PropertyType_GuidArray ||\
|
|
|
|
|
type == ABI::Windows::Foundation::PropertyType::PropertyType_PointArray || type == ABI::Windows::Foundation::PropertyType::PropertyType_SizeArray ||\
|
|
|
|
|
type == ABI::Windows::Foundation::PropertyType::PropertyType_RectArray || type == ABI::Windows::Foundation::PropertyType::PropertyType_OtherTypeArray)
|
|
|
|
|
|
|
|
|
|
template<typename _Type, bool bUnknown = std::is_base_of<IUnknown, _Type>::value>
|
|
|
|
|
struct winrt_type
|
|
|
|
|
{
|
|
|
|
|
};
|
|
|
|
|
template<typename _Type>
|
|
|
|
|
struct winrt_type<_Type, true>
|
|
|
|
|
{
|
|
|
|
|
static IUnknown* create(_Type* _ObjInCtx) {
|
|
|
|
|
return reinterpret_cast<IUnknown*>(_ObjInCtx);
|
|
|
|
|
}
|
|
|
|
|
static IID getuuid() { return __uuidof(_Type); }
|
|
|
|
|
static const ABI::Windows::Foundation::PropertyType _PropType = ABI::Windows::Foundation::PropertyType::PropertyType_OtherType;
|
|
|
|
|
};
|
|
|
|
|
template <typename _Type>
|
|
|
|
|
struct winrt_type<_Type, false>
|
|
|
|
|
{
|
|
|
|
|
static IUnknown* create(_Type* _ObjInCtx) {
|
|
|
|
|
Microsoft::WRL::ComPtr<IInspectable> _PObj;
|
|
|
|
|
Microsoft::WRL::ComPtr<IActivationFactory> objFactory;
|
|
|
|
|
HRESULT hr = Windows::Foundation::GetActivationFactory(Microsoft::WRL::Wrappers::HStringReference(RuntimeClass_Windows_Foundation_PropertyValue).Get(), objFactory.ReleaseAndGetAddressOf());
|
|
|
|
|
if (FAILED(hr)) return nullptr;
|
|
|
|
|
Microsoft::WRL::ComPtr<ABI::Windows::Foundation::IPropertyValueStatics> spPropVal;
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = objFactory.As(&spPropVal);
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = winrt_type<_Type>::create(spPropVal.Get(), _ObjInCtx, _PObj.GetAddressOf());
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
return reinterpret_cast<IUnknown*>(_PObj.Detach());
|
|
|
|
|
}
|
|
|
|
|
return nullptr;
|
|
|
|
|
}
|
|
|
|
|
static IID getuuid() { return __uuidof(ABI::Windows::Foundation::IPropertyValue); }
|
|
|
|
|
static const ABI::Windows::Foundation::PropertyType _PropType = ABI::Windows::Foundation::PropertyType::PropertyType_OtherType;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
template<>
|
|
|
|
|
struct winrt_type<void>
|
|
|
|
|
{
|
|
|
|
|
static HRESULT create(ABI::Windows::Foundation::IPropertyValueStatics* spPropVal, void* _ObjInCtx, IInspectable** ppInsp) {
|
|
|
|
|
(void)_ObjInCtx;
|
|
|
|
|
return spPropVal->CreateEmpty(ppInsp);
|
|
|
|
|
}
|
|
|
|
|
static const ABI::Windows::Foundation::PropertyType _PropType = ABI::Windows::Foundation::PropertyType::PropertyType_Empty;
|
|
|
|
|
};
|
|
|
|
|
#define MAKE_TYPE(Type, Name) template<>\
|
|
|
|
|
struct winrt_type<Type>\
|
|
|
|
|
{\
|
|
|
|
|
static HRESULT create(ABI::Windows::Foundation::IPropertyValueStatics* spPropVal, Type* _ObjInCtx, IInspectable** ppInsp) {\
|
|
|
|
|
return spPropVal->Create##Name(*_ObjInCtx, ppInsp);\
|
|
|
|
|
}\
|
|
|
|
|
static const ABI::Windows::Foundation::PropertyType _PropType = ABI::Windows::Foundation::PropertyType::PropertyType_##Name;\
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
template<typename _Type>
|
|
|
|
|
struct winrt_array_type
|
|
|
|
|
{
|
|
|
|
|
static IUnknown* create(_Type* _ObjInCtx, size_t N) {
|
|
|
|
|
Microsoft::WRL::ComPtr<IInspectable> _PObj;
|
|
|
|
|
Microsoft::WRL::ComPtr<IActivationFactory> objFactory;
|
|
|
|
|
HRESULT hr = Windows::Foundation::GetActivationFactory(Microsoft::WRL::Wrappers::HStringReference(RuntimeClass_Windows_Foundation_PropertyValue).Get(), objFactory.ReleaseAndGetAddressOf());
|
|
|
|
|
if (FAILED(hr)) return nullptr;
|
|
|
|
|
Microsoft::WRL::ComPtr<ABI::Windows::Foundation::IPropertyValueStatics> spPropVal;
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = objFactory.As(&spPropVal);
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = winrt_array_type<_Type>::create(spPropVal.Get(), N, _ObjInCtx, _PObj.GetAddressOf());
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
return reinterpret_cast<IUnknown*>(_PObj.Detach());
|
|
|
|
|
}
|
|
|
|
|
return nullptr;
|
|
|
|
|
}
|
|
|
|
|
static const ABI::Windows::Foundation::PropertyType _PropType = ABI::Windows::Foundation::PropertyType::PropertyType_OtherTypeArray;
|
|
|
|
|
};
|
|
|
|
|
template<int>
|
|
|
|
|
struct winrt_prop_type {};
|
|
|
|
|
|
|
|
|
|
template <>
|
|
|
|
|
struct winrt_prop_type<ABI::Windows::Foundation::PropertyType_Empty> {
|
|
|
|
|
typedef void _Type;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
template <>
|
|
|
|
|
struct winrt_prop_type<ABI::Windows::Foundation::PropertyType_OtherType> {
|
|
|
|
|
typedef void _Type;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
template <>
|
|
|
|
|
struct winrt_prop_type<ABI::Windows::Foundation::PropertyType_OtherTypeArray> {
|
|
|
|
|
typedef void _Type;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
#define MAKE_PROP(Prop, Type) template <>\
|
|
|
|
|
struct winrt_prop_type<ABI::Windows::Foundation::PropertyType_##Prop> {\
|
|
|
|
|
typedef Type _Type;\
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
#define MAKE_ARRAY_TYPE(Type, Name) MAKE_PROP(Name, Type)\
|
|
|
|
|
MAKE_PROP(Name##Array, Type*)\
|
|
|
|
|
MAKE_TYPE(Type, Name)\
|
|
|
|
|
template<>\
|
|
|
|
|
struct winrt_array_type<Type*>\
|
|
|
|
|
{\
|
|
|
|
|
static HRESULT create(ABI::Windows::Foundation::IPropertyValueStatics* spPropVal, UINT32 __valueSize, Type** _ObjInCtx, IInspectable** ppInsp) {\
|
|
|
|
|
return spPropVal->Create##Name##Array(__valueSize, *_ObjInCtx, ppInsp);\
|
|
|
|
|
}\
|
|
|
|
|
static const ABI::Windows::Foundation::PropertyType _PropType = ABI::Windows::Foundation::PropertyType::PropertyType_##Name##Array;\
|
|
|
|
|
static std::vector<Type> PropertyValueToVector(ABI::Windows::Foundation::IPropertyValue* propValue)\
|
|
|
|
|
{\
|
|
|
|
|
UINT32 uLen = 0;\
|
|
|
|
|
Type* pArray = nullptr;\
|
|
|
|
|
propValue->Get##Name##Array(&uLen, &pArray);\
|
|
|
|
|
return std::vector<Type>(pArray, pArray + uLen);\
|
|
|
|
|
}\
|
|
|
|
|
};
|
|
|
|
|
MAKE_ARRAY_TYPE(BYTE, UInt8)
|
|
|
|
|
MAKE_ARRAY_TYPE(INT16, Int16)
|
|
|
|
|
MAKE_ARRAY_TYPE(UINT16, UInt16)
|
|
|
|
|
MAKE_ARRAY_TYPE(INT32, Int32)
|
|
|
|
|
MAKE_ARRAY_TYPE(UINT32, UInt32)
|
|
|
|
|
MAKE_ARRAY_TYPE(INT64, Int64)
|
|
|
|
|
MAKE_ARRAY_TYPE(UINT64, UInt64)
|
|
|
|
|
MAKE_ARRAY_TYPE(FLOAT, Single)
|
|
|
|
|
MAKE_ARRAY_TYPE(DOUBLE, Double)
|
|
|
|
|
MAKE_ARRAY_TYPE(WCHAR, Char16)
|
|
|
|
|
//MAKE_ARRAY_TYPE(boolean, Boolean) //conflict with identical type in C++ of BYTE/UInt8
|
|
|
|
|
MAKE_ARRAY_TYPE(HSTRING, String)
|
|
|
|
|
MAKE_ARRAY_TYPE(IInspectable*, Inspectable)
|
|
|
|
|
MAKE_ARRAY_TYPE(GUID, Guid)
|
|
|
|
|
MAKE_ARRAY_TYPE(ABI::Windows::Foundation::DateTime, DateTime)
|
|
|
|
|
MAKE_ARRAY_TYPE(ABI::Windows::Foundation::TimeSpan, TimeSpan)
|
|
|
|
|
MAKE_ARRAY_TYPE(ABI::Windows::Foundation::Point, Point)
|
|
|
|
|
MAKE_ARRAY_TYPE(ABI::Windows::Foundation::Size, Size)
|
|
|
|
|
MAKE_ARRAY_TYPE(ABI::Windows::Foundation::Rect, Rect)
|
|
|
|
|
|
|
|
|
|
template < typename T >
|
|
|
|
|
struct DerefHelper
|
|
|
|
|
{
|
|
|
|
|
typedef T DerefType;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
template < typename T >
|
|
|
|
|
struct DerefHelper<T*>
|
|
|
|
|
{
|
|
|
|
|
typedef T DerefType;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
#define __is_valid_winrt_type(_Type) (std::is_void<_Type>::value || \
|
|
|
|
|
std::is_same<_Type, BYTE>::value || \
|
|
|
|
|
std::is_same<_Type, INT16>::value || \
|
|
|
|
|
std::is_same<_Type, UINT16>::value || \
|
|
|
|
|
std::is_same<_Type, INT32>::value || \
|
|
|
|
|
std::is_same<_Type, UINT32>::value || \
|
|
|
|
|
std::is_same<_Type, INT64>::value || \
|
|
|
|
|
std::is_same<_Type, UINT64>::value || \
|
|
|
|
|
std::is_same<_Type, FLOAT>::value || \
|
|
|
|
|
std::is_same<_Type, DOUBLE>::value || \
|
|
|
|
|
std::is_same<_Type, WCHAR>::value || \
|
|
|
|
|
std::is_same<_Type, boolean>::value || \
|
|
|
|
|
std::is_same<_Type, HSTRING>::value || \
|
|
|
|
|
std::is_same<_Type, IInspectable *>::value || \
|
|
|
|
|
std::is_base_of<Microsoft::WRL::Details::RuntimeClassBase, _Type>::value || \
|
|
|
|
|
std::is_base_of<IInspectable, typename DerefHelper<_Type>::DerefType>::value || \
|
|
|
|
|
std::is_same<_Type, GUID>::value || \
|
|
|
|
|
std::is_same<_Type, ABI::Windows::Foundation::DateTime>::value || \
|
|
|
|
|
std::is_same<_Type, ABI::Windows::Foundation::TimeSpan>::value || \
|
|
|
|
|
std::is_same<_Type, ABI::Windows::Foundation::Point>::value || \
|
|
|
|
|
std::is_same<_Type, ABI::Windows::Foundation::Size>::value || \
|
|
|
|
|
std::is_same<_Type, ABI::Windows::Foundation::Rect>::value || \
|
|
|
|
|
std::is_same<_Type, BYTE*>::value || \
|
|
|
|
|
std::is_same<_Type, INT16*>::value || \
|
|
|
|
|
std::is_same<_Type, UINT16*>::value || \
|
|
|
|
|
std::is_same<_Type, INT32*>::value || \
|
|
|
|
|
std::is_same<_Type, UINT32*>::value || \
|
|
|
|
|
std::is_same<_Type, INT64*>::value || \
|
|
|
|
|
std::is_same<_Type, UINT64*>::value || \
|
|
|
|
|
std::is_same<_Type, FLOAT*>::value || \
|
|
|
|
|
std::is_same<_Type, DOUBLE*>::value || \
|
|
|
|
|
std::is_same<_Type, WCHAR*>::value || \
|
|
|
|
|
std::is_same<_Type, boolean*>::value || \
|
|
|
|
|
std::is_same<_Type, HSTRING*>::value || \
|
|
|
|
|
std::is_same<_Type, IInspectable **>::value || \
|
|
|
|
|
std::is_same<_Type, GUID*>::value || \
|
|
|
|
|
std::is_same<_Type, ABI::Windows::Foundation::DateTime*>::value || \
|
|
|
|
|
std::is_same<_Type, ABI::Windows::Foundation::TimeSpan*>::value || \
|
|
|
|
|
std::is_same<_Type, ABI::Windows::Foundation::Point*>::value || \
|
|
|
|
|
std::is_same<_Type, ABI::Windows::Foundation::Size*>::value || \
|
|
|
|
|
std::is_same<_Type, ABI::Windows::Foundation::Rect*>::value)
|
|
|
|
|
#endif
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#else
|
|
|
|
|
EXTERN_C const IID IID_ICustomStreamSink;
|
|
|
|
|
|
|
|
|
|
class DECLSPEC_UUID("4F8A1939-2FD3-46DB-AE70-DB7E0DD79B73") DECLSPEC_NOVTABLE ICustomStreamSink : public IUnknown
|
|
|
|
|
{
|
|
|
|
|
public:
|
|
|
|
|
virtual HRESULT Initialize() = 0;
|
|
|
|
|
virtual HRESULT Shutdown() = 0;
|
|
|
|
|
virtual HRESULT Start(MFTIME start) = 0;
|
|
|
|
|
virtual HRESULT Pause() = 0;
|
|
|
|
|
virtual HRESULT Restart() = 0;
|
|
|
|
|
virtual HRESULT Stop() = 0;
|
|
|
|
|
};
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
#define MF_PROP_SAMPLEGRABBERCALLBACK L"samplegrabbercallback"
|
|
|
|
|
#define MF_PROP_VIDTYPE L"vidtype"
|
|
|
|
|
#define MF_PROP_VIDENCPROPS L"videncprops"
|
|
|
|
|
|
|
|
|
|
#include <initguid.h>
|
|
|
|
|
|
|
|
|
|
// MF_MEDIASINK_SAMPLEGRABBERCALLBACK: {26957AA7-AFF4-464c-BB8B-07BA65CE11DF}
|
|
|
|
|
// Type: IUnknown*
|
|
|
|
|
DEFINE_GUID(MF_MEDIASINK_SAMPLEGRABBERCALLBACK,
|
|
|
|
|
0x26957aa7, 0xaff4, 0x464c, 0xbb, 0x8b, 0x7, 0xba, 0x65, 0xce, 0x11, 0xdf);
|
|
|
|
|
|
|
|
|
|
// {4BD133CC-EB9B-496E-8865-0813BFBC6FAA}
|
|
|
|
|
DEFINE_GUID(MF_STREAMSINK_ID, 0x4bd133cc, 0xeb9b, 0x496e, 0x88, 0x65, 0x8, 0x13, 0xbf, 0xbc, 0x6f, 0xaa);
|
|
|
|
|
|
|
|
|
|
// {C9E22A8C-6A50-4D78-9183-0834A02A3780}
|
|
|
|
|
DEFINE_GUID(MF_STREAMSINK_MEDIASINKINTERFACE,
|
|
|
|
|
0xc9e22a8c, 0x6a50, 0x4d78, 0x91, 0x83, 0x8, 0x34, 0xa0, 0x2a, 0x37, 0x80);
|
|
|
|
|
|
|
|
|
|
// {DABD13AB-26B7-47C2-97C1-4B04C187B838}
|
|
|
|
|
DEFINE_GUID(MF_MEDIASINK_PREFERREDTYPE,
|
|
|
|
|
0xdabd13ab, 0x26b7, 0x47c2, 0x97, 0xc1, 0x4b, 0x4, 0xc1, 0x87, 0xb8, 0x38);
|
|
|
|
|
|
|
|
|
|
#include <utility>
|
|
|
|
|
#ifdef _UNICODE
|
|
|
|
|
#define MAKE_MAP(e) std::map<e, std::wstring>
|
|
|
|
|
#define MAKE_ENUM(e) std::pair<e, std::wstring>
|
|
|
|
|
#define MAKE_ENUM_PAIR(e, str) std::pair<e, std::wstring>(str, L#str)
|
|
|
|
|
#else
|
|
|
|
|
#define MAKE_MAP(e) std::map<e, std::string>
|
|
|
|
|
#define MAKE_ENUM(e) std::pair<e, std::string>
|
|
|
|
|
#define MAKE_ENUM_PAIR(e, str) std::pair<e, std::string>(str, #str)
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
MAKE_ENUM(MediaEventType) MediaEventTypePairs[] = {
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEUnknown),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEError),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEExtendedType),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MENonFatalError),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEGenericV1Anchor),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESessionUnknown),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESessionTopologySet),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESessionTopologiesCleared),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESessionStarted),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESessionPaused),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESessionStopped),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESessionClosed),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESessionEnded),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESessionRateChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESessionScrubSampleComplete),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESessionCapabilitiesChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESessionTopologyStatus),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESessionNotifyPresentationTime),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MENewPresentation),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MELicenseAcquisitionStart),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MELicenseAcquisitionCompleted),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEIndividualizationStart),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEIndividualizationCompleted),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEEnablerProgress),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEEnablerCompleted),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEPolicyError),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEPolicyReport),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEBufferingStarted),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEBufferingStopped),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEConnectStart),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEConnectEnd),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEReconnectStart),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEReconnectEnd),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MERendererEvent),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESessionStreamSinkFormatChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESessionV1Anchor),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESourceUnknown),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESourceStarted),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamStarted),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESourceSeeked),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamSeeked),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MENewStream),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEUpdatedStream),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESourceStopped),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamStopped),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESourcePaused),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamPaused),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEEndOfPresentation),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEEndOfStream),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEMediaSample),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamTick),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamThinMode),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamFormatChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESourceRateChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEEndOfPresentationSegment),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESourceCharacteristicsChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESourceRateChangeRequested),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESourceMetadataChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESequencerSourceTopologyUpdated),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESourceV1Anchor),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESinkUnknown),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamSinkStarted),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamSinkStopped),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamSinkPaused),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamSinkRateChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamSinkRequestSample),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamSinkMarker),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamSinkPrerolled),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamSinkScrubSampleComplete),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamSinkFormatChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEStreamSinkDeviceChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEQualityNotify),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESinkInvalidated),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEAudioSessionNameChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEAudioSessionVolumeChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEAudioSessionDeviceRemoved),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEAudioSessionServerShutdown),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEAudioSessionGroupingParamChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEAudioSessionIconChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEAudioSessionFormatChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEAudioSessionDisconnected),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEAudioSessionExclusiveModeOverride),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESinkV1Anchor),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MECaptureAudioSessionVolumeChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MECaptureAudioSessionDeviceRemoved),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MECaptureAudioSessionFormatChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MECaptureAudioSessionDisconnected),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MECaptureAudioSessionExclusiveModeOverride),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MECaptureAudioSessionServerShutdown),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MESinkV2Anchor),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, METrustUnknown),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEPolicyChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEContentProtectionMessage),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEPolicySet),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, METrustV1Anchor),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEWMDRMLicenseBackupCompleted),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEWMDRMLicenseBackupProgress),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEWMDRMLicenseRestoreCompleted),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEWMDRMLicenseRestoreProgress),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEWMDRMLicenseAcquisitionCompleted),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEWMDRMIndividualizationCompleted),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEWMDRMIndividualizationProgress),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEWMDRMProximityCompleted),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEWMDRMLicenseStoreCleaned),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEWMDRMRevocationDownloadCompleted),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEWMDRMV1Anchor),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, METransformUnknown),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, METransformNeedInput),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, METransformHaveOutput),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, METransformDrainComplete),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, METransformMarker),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEByteStreamCharacteristicsChanged),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEVideoCaptureDeviceRemoved),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEVideoCaptureDevicePreempted),
|
|
|
|
|
MAKE_ENUM_PAIR(MediaEventType, MEReservedMax)
|
|
|
|
|
};
|
|
|
|
|
MAKE_MAP(MediaEventType) MediaEventTypeMap(MediaEventTypePairs, MediaEventTypePairs + sizeof(MediaEventTypePairs) / sizeof(MediaEventTypePairs[0]));
|
|
|
|
|
|
|
|
|
|
MAKE_ENUM(MFSTREAMSINK_MARKER_TYPE) StreamSinkMarkerTypePairs[] = {
|
|
|
|
|
MAKE_ENUM_PAIR(MFSTREAMSINK_MARKER_TYPE, MFSTREAMSINK_MARKER_DEFAULT),
|
|
|
|
|
MAKE_ENUM_PAIR(MFSTREAMSINK_MARKER_TYPE, MFSTREAMSINK_MARKER_ENDOFSEGMENT),
|
|
|
|
|
MAKE_ENUM_PAIR(MFSTREAMSINK_MARKER_TYPE, MFSTREAMSINK_MARKER_TICK),
|
|
|
|
|
MAKE_ENUM_PAIR(MFSTREAMSINK_MARKER_TYPE, MFSTREAMSINK_MARKER_EVENT)
|
|
|
|
|
};
|
|
|
|
|
MAKE_MAP(MFSTREAMSINK_MARKER_TYPE) StreamSinkMarkerTypeMap(StreamSinkMarkerTypePairs, StreamSinkMarkerTypePairs + sizeof(StreamSinkMarkerTypePairs) / sizeof(StreamSinkMarkerTypePairs[0]));
|
|
|
|
|
|
|
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#ifdef __cplusplus_winrt
|
|
|
|
|
#define _ContextCallback Concurrency::details::_ContextCallback
|
|
|
|
|
#define BEGIN_CALL_IN_CONTEXT(hr, var, ...) hr = S_OK;\
|
|
|
|
|
var._CallInContext([__VA_ARGS__]() {
|
|
|
|
|
#define END_CALL_IN_CONTEXT(hr) if (FAILED(hr)) throw Platform::Exception::CreateException(hr);\
|
|
|
|
|
});
|
|
|
|
|
#define END_CALL_IN_CONTEXT_BASE });
|
|
|
|
|
#else
|
|
|
|
|
#define _ContextCallback Concurrency_winrt::details::_ContextCallback
|
|
|
|
|
#define BEGIN_CALL_IN_CONTEXT(hr, var, ...) hr = var._CallInContext([__VA_ARGS__]() -> HRESULT {
|
|
|
|
|
#define END_CALL_IN_CONTEXT(hr) return hr;\
|
|
|
|
|
});
|
|
|
|
|
#define END_CALL_IN_CONTEXT_BASE return S_OK;\
|
|
|
|
|
});
|
|
|
|
|
#endif
|
|
|
|
|
#define GET_CURRENT_CONTEXT _ContextCallback::_CaptureCurrent()
|
|
|
|
|
#define SAVE_CURRENT_CONTEXT(var) _ContextCallback var = GET_CURRENT_CONTEXT
|
|
|
|
|
|
|
|
|
|
#define COMMA ,
|
|
|
|
|
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#ifdef __cplusplus_winrt
|
|
|
|
|
#define _Object Platform::Object^
|
|
|
|
|
#define _ObjectObj Platform::Object^
|
|
|
|
|
#define _String Platform::String^
|
|
|
|
|
#define _StringObj Platform::String^
|
|
|
|
|
#define _StringReference ref new Platform::String
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define _StringReferenceObj Platform::String^
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define _DeviceInformationCollection Windows::Devices::Enumeration::DeviceInformationCollection
|
|
|
|
|
#define _MediaCapture Windows::Media::Capture::MediaCapture
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
#define _MediaCaptureVideoPreview Windows::Media::Capture::MediaCapture
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define _MediaCaptureInitializationSettings Windows::Media::Capture::MediaCaptureInitializationSettings
|
|
|
|
|
#define _VideoDeviceController Windows::Media::Devices::VideoDeviceController
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
#define _MediaDeviceController Windows::Media::Devices::VideoDeviceController
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define _MediaEncodingProperties Windows::Media::MediaProperties::IMediaEncodingProperties
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
#define _VideoEncodingProperties Windows::Media::MediaProperties::VideoEncodingProperties
|
|
|
|
|
#define _MediaStreamType Windows::Media::Capture::MediaStreamType
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define _AsyncInfo Windows::Foundation::IAsyncInfo
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define _AsyncAction Windows::Foundation::IAsyncAction
|
|
|
|
|
#define _AsyncOperation Windows::Foundation::IAsyncOperation
|
|
|
|
|
#define _DeviceClass Windows::Devices::Enumeration::DeviceClass
|
|
|
|
|
#define _IDeviceInformation Windows::Devices::Enumeration::DeviceInformation
|
|
|
|
|
#define _DeviceInformation Windows::Devices::Enumeration::DeviceInformation
|
|
|
|
|
#define _DeviceInformationStatics Windows::Devices::Enumeration::DeviceInformation
|
|
|
|
|
#define _MediaEncodingProfile Windows::Media::MediaProperties::MediaEncodingProfile
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
#define _StreamingCaptureMode Windows::Media::Capture::StreamingCaptureMode
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define _PropertySet Windows::Foundation::Collections::PropertySet
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
#define _Map Windows::Foundation::Collections::PropertySet
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define _PropertyValueStatics Windows::Foundation::PropertyValue
|
|
|
|
|
#define _VectorView Windows::Foundation::Collections::IVectorView
|
|
|
|
|
#define _StartPreviewToCustomSinkIdAsync StartPreviewToCustomSinkAsync
|
|
|
|
|
#define _InitializeWithSettingsAsync InitializeAsync
|
|
|
|
|
#define _FindAllAsyncDeviceClass FindAllAsync
|
|
|
|
|
#define _MediaExtension Windows::Media::IMediaExtension
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define BEGIN_CREATE_ASYNC(type, ...) (Concurrency::create_async([__VA_ARGS__]() {
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define END_CREATE_ASYNC(hr) if (FAILED(hr)) throw Platform::Exception::CreateException(hr);\
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
}))
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define DEFINE_TASK Concurrency::task
|
|
|
|
|
#define CREATE_TASK Concurrency::create_task
|
|
|
|
|
#define CREATE_OR_CONTINUE_TASK(_task, rettype, func) _task = (_task == Concurrency::task<rettype>()) ? Concurrency::create_task(func) : _task.then([func](rettype) -> rettype { return func(); });
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define DEFINE_RET_VAL(x)
|
|
|
|
|
#define DEFINE_RET_TYPE(x)
|
|
|
|
|
#define DEFINE_RET_FORMAL(x) x
|
|
|
|
|
#define RET_VAL(x) return x;
|
|
|
|
|
#define RET_VAL_BASE
|
|
|
|
|
#define MAKE_STRING(str) str
|
|
|
|
|
#define GET_STL_STRING(str) std::wstring(str->Data())
|
|
|
|
|
#define GET_STL_STRING_RAW(str) std::wstring(str->Data())
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define MAKE_WRL_OBJ(x) x^
|
|
|
|
|
#define MAKE_WRL_REF(x) x^
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define MAKE_OBJ_REF(x) x^
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define MAKE_WRL_AGILE_REF(x) Platform::Agile<x^>
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define MAKE_PROPERTY_BACKING(Type, PropName) property Type PropName;
|
|
|
|
|
#define MAKE_PROPERTY(Type, PropName, PropValue)
|
|
|
|
|
#define MAKE_PROPERTY_STRING(Type, PropName, PropValue)
|
|
|
|
|
#define MAKE_READONLY_PROPERTY(Type, PropName, PropValue) property Type PropName\
|
|
|
|
|
{\
|
|
|
|
|
Type get() { return PropValue; }\
|
|
|
|
|
}
|
|
|
|
|
#define THROW_INVALID_ARG throw ref new Platform::InvalidArgumentException();
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define RELEASE_AGILE_WRL(x) x = nullptr;
|
|
|
|
|
#define RELEASE_WRL(x) x = nullptr;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
#define GET_WRL_OBJ_FROM_REF(objtype, obj, orig, hr) objtype^ obj = orig;\
|
|
|
|
|
hr = S_OK;
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define GET_WRL_OBJ_FROM_OBJ(objtype, obj, orig, hr) objtype^ obj = safe_cast<objtype^>(orig);\
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
hr = S_OK;
|
|
|
|
|
#define WRL_ENUM_GET(obj, prefix, prop) obj::##prop
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define WRL_PROP_GET(obj, prop, arg, hr) arg = obj->##prop;\
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
hr = S_OK;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define WRL_PROP_PUT(obj, prop, arg, hr) obj->##prop = arg;\
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
hr = S_OK;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define WRL_METHOD_BASE(obj, method, ret, hr) ret = obj->##method();\
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
hr = S_OK;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define WRL_METHOD(obj, method, ret, hr, ...) ret = obj->##method(__VA_ARGS__);\
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
hr = S_OK;
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define WRL_METHOD_NORET_BASE(obj, method, hr) obj->##method();\
|
|
|
|
|
hr = S_OK;
|
|
|
|
|
#define WRL_METHOD_NORET(obj, method, hr, ...) obj->##method(__VA_ARGS__);\
|
|
|
|
|
hr = S_OK;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define REF_WRL_OBJ(obj) &obj
|
|
|
|
|
#define DEREF_WRL_OBJ(obj) obj
|
|
|
|
|
#define DEREF_AGILE_WRL_OBJ(obj) obj.Get()
|
|
|
|
|
#define DEREF_AS_NATIVE_WRL_OBJ(type, obj) reinterpret_cast<type*>(obj)
|
|
|
|
|
#define PREPARE_TRANSFER_WRL_OBJ(obj) obj
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define ACTIVATE_LOCAL_OBJ_BASE(objtype) ref new objtype()
|
|
|
|
|
#define ACTIVATE_LOCAL_OBJ(objtype, ...) ref new objtype(__VA_ARGS__)
|
|
|
|
|
#define ACTIVATE_EVENT_HANDLER(objtype, ...) ref new objtype(__VA_ARGS__)
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
#define ACTIVATE_OBJ(rtclass, objtype, obj, hr) MAKE_WRL_OBJ(objtype) obj = ref new objtype();\
|
|
|
|
|
hr = S_OK;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define ACTIVATE_STATIC_OBJ(rtclass, objtype, obj, hr) objtype obj;\
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
hr = S_OK;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#else
|
|
|
|
|
#define _Object IInspectable*
|
|
|
|
|
#define _ObjectObj Microsoft::WRL::ComPtr<IInspectable>
|
|
|
|
|
#define _String HSTRING
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
#define _StringObj Microsoft::WRL::Wrappers::HString
|
|
|
|
|
#define _StringReference Microsoft::WRL::Wrappers::HStringReference
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define _StringReferenceObj Microsoft::WRL::Wrappers::HStringReference
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define _DeviceInformationCollection ABI::Windows::Devices::Enumeration::DeviceInformationCollection
|
|
|
|
|
#define _MediaCapture ABI::Windows::Media::Capture::IMediaCapture
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
#define _MediaCaptureVideoPreview ABI::Windows::Media::Capture::IMediaCaptureVideoPreview
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define _MediaCaptureInitializationSettings ABI::Windows::Media::Capture::IMediaCaptureInitializationSettings
|
|
|
|
|
#define _VideoDeviceController ABI::Windows::Media::Devices::IVideoDeviceController
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
#define _MediaDeviceController ABI::Windows::Media::Devices::IMediaDeviceController
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define _MediaEncodingProperties ABI::Windows::Media::MediaProperties::IMediaEncodingProperties
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
#define _VideoEncodingProperties ABI::Windows::Media::MediaProperties::IVideoEncodingProperties
|
|
|
|
|
#define _MediaStreamType ABI::Windows::Media::Capture::MediaStreamType
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define _AsyncInfo ABI::Windows::Foundation::IAsyncInfo
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define _AsyncAction ABI::Windows::Foundation::IAsyncAction
|
|
|
|
|
#define _AsyncOperation ABI::Windows::Foundation::IAsyncOperation
|
|
|
|
|
#define _DeviceClass ABI::Windows::Devices::Enumeration::DeviceClass
|
|
|
|
|
#define _IDeviceInformation ABI::Windows::Devices::Enumeration::IDeviceInformation
|
|
|
|
|
#define _DeviceInformation ABI::Windows::Devices::Enumeration::DeviceInformation
|
|
|
|
|
#define _DeviceInformationStatics ABI::Windows::Devices::Enumeration::IDeviceInformationStatics
|
|
|
|
|
#define _MediaEncodingProfile ABI::Windows::Media::MediaProperties::IMediaEncodingProfile
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
#define _StreamingCaptureMode ABI::Windows::Media::Capture::StreamingCaptureMode
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define _PropertySet ABI::Windows::Foundation::Collections::IPropertySet
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
#define _Map ABI::Windows::Foundation::Collections::IMap<HSTRING, IInspectable *>
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define _PropertyValueStatics ABI::Windows::Foundation::IPropertyValueStatics
|
|
|
|
|
#define _VectorView ABI::Windows::Foundation::Collections::IVectorView
|
|
|
|
|
#define _StartPreviewToCustomSinkIdAsync StartPreviewToCustomSinkIdAsync
|
|
|
|
|
#define _InitializeWithSettingsAsync InitializeWithSettingsAsync
|
|
|
|
|
#define _FindAllAsyncDeviceClass FindAllAsyncDeviceClass
|
|
|
|
|
#define _MediaExtension ABI::Windows::Media::IMediaExtension
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define BEGIN_CREATE_ASYNC(type, ...) Concurrency_winrt::create_async<type>([__VA_ARGS__]() -> HRESULT {
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define END_CREATE_ASYNC(hr) return hr;\
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
})
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define DEFINE_TASK Concurrency_winrt::task
|
|
|
|
|
#define CREATE_TASK Concurrency_winrt::create_task
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define CREATE_OR_CONTINUE_TASK(_task, rettype, func) _task = (_task == Concurrency_winrt::task<rettype>()) ? Concurrency_winrt::create_task<rettype>(func) : _task.then([func](rettype, rettype* retVal) -> HRESULT { return func(retVal); });
|
|
|
|
|
#define DEFINE_RET_VAL(x) x* retVal
|
|
|
|
|
#define DEFINE_RET_TYPE(x) <x>
|
|
|
|
|
#define DEFINE_RET_FORMAL(x) HRESULT
|
|
|
|
|
#define RET_VAL(x) *retVal = x;\
|
|
|
|
|
return S_OK;
|
|
|
|
|
#define RET_VAL_BASE return S_OK;
|
|
|
|
|
#define MAKE_STRING(str) Microsoft::WRL::Wrappers::HStringReference(L##str)
|
|
|
|
|
#define GET_STL_STRING(str) std::wstring(str.GetRawBuffer(NULL))
|
|
|
|
|
#define GET_STL_STRING_RAW(str) WindowsGetStringRawBuffer(str, NULL)
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define MAKE_WRL_OBJ(x) Microsoft::WRL::ComPtr<x>
|
|
|
|
|
#define MAKE_WRL_REF(x) x*
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define MAKE_OBJ_REF(x) x
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define MAKE_WRL_AGILE_REF(x) x*
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define MAKE_PROPERTY_BACKING(Type, PropName) Type PropName;
|
|
|
|
|
#define MAKE_PROPERTY(Type, PropName, PropValue) STDMETHODIMP get_##PropName(Type* pVal) { if (pVal) { *pVal = PropValue; } else { return E_INVALIDARG; } return S_OK; }\
|
|
|
|
|
STDMETHODIMP put_##PropName(Type Val) { PropValue = Val; return S_OK; }
|
|
|
|
|
#define MAKE_PROPERTY_STRING(Type, PropName, PropValue) STDMETHODIMP get_##PropName(Type* pVal) { if (pVal) { return ::WindowsDuplicateString(PropValue.Get(), pVal); } else { return E_INVALIDARG; } }\
|
|
|
|
|
STDMETHODIMP put_##PropName(Type Val) { return PropValue.Set(Val); }
|
|
|
|
|
#define MAKE_READONLY_PROPERTY(Type, PropName, PropValue) STDMETHODIMP get_##PropName(Type* pVal) { if (pVal) { *pVal = PropValue; } else { return E_INVALIDARG; } return S_OK; }
|
|
|
|
|
#define THROW_INVALID_ARG RoOriginateError(E_INVALIDARG, nullptr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define RELEASE_AGILE_WRL(x) if (x) { (x)->Release(); x = nullptr; }
|
|
|
|
|
#define RELEASE_WRL(x) if (x) { (x)->Release(); x = nullptr; }
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
#define GET_WRL_OBJ_FROM_REF(objtype, obj, orig, hr) Microsoft::WRL::ComPtr<objtype> obj;\
|
|
|
|
|
hr = orig->QueryInterface(__uuidof(objtype), &obj);
|
|
|
|
|
#define GET_WRL_OBJ_FROM_OBJ(objtype, obj, orig, hr) Microsoft::WRL::ComPtr<objtype> obj;\
|
|
|
|
|
hr = orig.As(&obj);
|
|
|
|
|
#define WRL_ENUM_GET(obj, prefix, prop) obj::prefix##_##prop
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define WRL_PROP_GET(obj, prop, arg, hr) hr = obj->get_##prop(&arg);
|
|
|
|
|
#define WRL_PROP_PUT(obj, prop, arg, hr) hr = obj->put_##prop(arg);
|
|
|
|
|
#define WRL_METHOD_BASE(obj, method, ret, hr) hr = obj->##method(&ret);
|
|
|
|
|
#define WRL_METHOD(obj, method, ret, hr, ...) hr = obj->##method(__VA_ARGS__, &ret);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define WRL_METHOD_NORET_BASE(obj, method, hr) hr = obj->##method();
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#define REF_WRL_OBJ(obj) obj.GetAddressOf()
|
|
|
|
|
#define DEREF_WRL_OBJ(obj) obj.Get()
|
|
|
|
|
#define DEREF_AGILE_WRL_OBJ(obj) obj
|
|
|
|
|
#define DEREF_AS_NATIVE_WRL_OBJ(type, obj) obj.Get()
|
|
|
|
|
#define PREPARE_TRANSFER_WRL_OBJ(obj) obj.Detach()
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define ACTIVATE_LOCAL_OBJ_BASE(objtype) Microsoft::WRL::Make<objtype>()
|
|
|
|
|
#define ACTIVATE_LOCAL_OBJ(objtype, ...) Microsoft::WRL::Make<objtype>(__VA_ARGS__)
|
|
|
|
|
#define ACTIVATE_EVENT_HANDLER(objtype, ...) Microsoft::WRL::Callback<objtype>(__VA_ARGS__).Get()
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
#define ACTIVATE_OBJ(rtclass, objtype, obj, hr) MAKE_WRL_OBJ(objtype) obj;\
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
{\
|
|
|
|
|
Microsoft::WRL::ComPtr<IActivationFactory> objFactory;\
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
hr = Windows::Foundation::GetActivationFactory(Microsoft::WRL::Wrappers::HStringReference(rtclass).Get(), objFactory.ReleaseAndGetAddressOf());\
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
if (SUCCEEDED(hr)) {\
|
|
|
|
|
Microsoft::WRL::ComPtr<IInspectable> pInsp;\
|
|
|
|
|
hr = objFactory->ActivateInstance(pInsp.GetAddressOf());\
|
|
|
|
|
if (SUCCEEDED(hr)) hr = pInsp.As(&obj);\
|
|
|
|
|
}\
|
|
|
|
|
}
|
|
|
|
|
#define ACTIVATE_STATIC_OBJ(rtclass, objtype, obj, hr) objtype obj;\
|
|
|
|
|
{\
|
|
|
|
|
Microsoft::WRL::ComPtr<IActivationFactory> objFactory;\
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
Support VS 2013 and consistency cleanup and C++/CX object creation fixed
11 years ago
|
|
|
hr = Windows::Foundation::GetActivationFactory(Microsoft::WRL::Wrappers::HStringReference(rtclass).Get(), objFactory.ReleaseAndGetAddressOf());\
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
if (SUCCEEDED(hr)) {\
|
|
|
|
|
if (SUCCEEDED(hr)) hr = objFactory.As(&obj);\
|
|
|
|
|
}\
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define _ComPtr Microsoft::WRL::ComPtr
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#else
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
|
|
|
|
|
template <class T>
|
|
|
|
|
class ComPtr : public ATL::CComPtr<T>
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
{
|
|
|
|
|
public:
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
ComPtr() throw()
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
{
|
|
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
ComPtr(int nNull) throw() :
|
|
|
|
|
CComPtr<T>((T*)nNull)
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
{
|
|
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
ComPtr(T* lp) throw() :
|
|
|
|
|
CComPtr<T>(lp)
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
|
|
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
}
|
|
|
|
|
ComPtr(_In_ const CComPtr<T>& lp) throw() :
|
|
|
|
|
CComPtr<T>(lp.p)
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
{
|
|
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
virtual ~ComPtr() {}
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
T* const* GetAddressOf() const throw()
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
return &p;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
|
|
|
|
|
T** GetAddressOf() throw()
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
return &p;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
|
|
|
|
|
T** ReleaseAndGetAddressOf() throw()
|
|
|
|
|
{
|
|
|
|
|
InternalRelease();
|
|
|
|
|
return &p;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
|
|
|
|
|
T* Get() const throw()
|
|
|
|
|
{
|
|
|
|
|
return p;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
ComPtr& operator=(decltype(__nullptr)) throw()
|
|
|
|
|
{
|
|
|
|
|
InternalRelease();
|
|
|
|
|
return *this;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
ComPtr& operator=(_In_ const int nNull) throw()
|
|
|
|
|
{
|
|
|
|
|
ASSERT(nNull == 0);
|
|
|
|
|
(void)nNull;
|
|
|
|
|
InternalRelease();
|
|
|
|
|
return *this;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
unsigned long Reset()
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
return InternalRelease();
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
// query for U interface
|
|
|
|
|
template<typename U>
|
|
|
|
|
HRESULT As(_Inout_ U** lp) const throw()
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
return p->QueryInterface(__uuidof(U), (void**)lp);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
// query for U interface
|
|
|
|
|
template<typename U>
|
|
|
|
|
HRESULT As(_Out_ ComPtr<U>* lp) const throw()
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
return p->QueryInterface(__uuidof(U), reinterpret_cast<void**>(lp->ReleaseAndGetAddressOf()));
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
|
|
|
|
private:
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
unsigned long InternalRelease() throw()
|
|
|
|
|
{
|
|
|
|
|
unsigned long ref = 0;
|
|
|
|
|
T* temp = p;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
if (temp != nullptr)
|
|
|
|
|
{
|
|
|
|
|
p = nullptr;
|
|
|
|
|
ref = temp->Release();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return ref;
|
|
|
|
|
}
|
|
|
|
|
};
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
#define _ComPtr ComPtr
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
template <class TBase=IMFAttributes>
|
|
|
|
|
class CBaseAttributes : public TBase
|
|
|
|
|
{
|
|
|
|
|
protected:
|
|
|
|
|
// This version of the constructor does not initialize the
|
|
|
|
|
// attribute store. The derived class must call Initialize() in
|
|
|
|
|
// its own constructor.
|
|
|
|
|
CBaseAttributes()
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// This version of the constructor initializes the attribute
|
|
|
|
|
// store, but the derived class must pass an HRESULT parameter
|
|
|
|
|
// to the constructor.
|
|
|
|
|
|
|
|
|
|
CBaseAttributes(HRESULT& hr, UINT32 cInitialSize = 0)
|
|
|
|
|
{
|
|
|
|
|
hr = Initialize(cInitialSize);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// The next version of the constructor uses a caller-provided
|
|
|
|
|
// implementation of IMFAttributes.
|
|
|
|
|
|
|
|
|
|
// (Sometimes you want to delegate IMFAttributes calls to some
|
|
|
|
|
// other object that implements IMFAttributes, rather than using
|
|
|
|
|
// MFCreateAttributes.)
|
|
|
|
|
|
|
|
|
|
CBaseAttributes(HRESULT& hr, IUnknown *pUnk)
|
|
|
|
|
{
|
|
|
|
|
hr = Initialize(pUnk);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
virtual ~CBaseAttributes()
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Initializes the object by creating the standard Media Foundation attribute store.
|
|
|
|
|
HRESULT Initialize(UINT32 cInitialSize = 0)
|
|
|
|
|
{
|
|
|
|
|
if (_spAttributes.Get() == nullptr)
|
|
|
|
|
{
|
|
|
|
|
return MFCreateAttributes(&_spAttributes, cInitialSize);
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
return S_OK;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Initializes this object from a caller-provided attribute store.
|
|
|
|
|
// pUnk: Pointer to an object that exposes IMFAttributes.
|
|
|
|
|
HRESULT Initialize(IUnknown *pUnk)
|
|
|
|
|
{
|
|
|
|
|
if (_spAttributes)
|
|
|
|
|
{
|
|
|
|
|
_spAttributes.Reset();
|
|
|
|
|
_spAttributes = nullptr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
return pUnk->QueryInterface(IID_PPV_ARGS(&_spAttributes));
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
public:
|
|
|
|
|
|
|
|
|
|
// IMFAttributes methods
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP GetItem(REFGUID guidKey, PROPVARIANT* pValue)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->GetItem(guidKey, pValue);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP GetItemType(REFGUID guidKey, MF_ATTRIBUTE_TYPE* pType)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->GetItemType(guidKey, pType);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP CompareItem(REFGUID guidKey, REFPROPVARIANT Value, BOOL* pbResult)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->CompareItem(guidKey, Value, pbResult);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP Compare(
|
|
|
|
|
IMFAttributes* pTheirs,
|
|
|
|
|
MF_ATTRIBUTES_MATCH_TYPE MatchType,
|
|
|
|
|
BOOL* pbResult
|
|
|
|
|
)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->Compare(pTheirs, MatchType, pbResult);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP GetUINT32(REFGUID guidKey, UINT32* punValue)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->GetUINT32(guidKey, punValue);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP GetUINT64(REFGUID guidKey, UINT64* punValue)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->GetUINT64(guidKey, punValue);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP GetDouble(REFGUID guidKey, double* pfValue)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->GetDouble(guidKey, pfValue);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP GetGUID(REFGUID guidKey, GUID* pguidValue)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->GetGUID(guidKey, pguidValue);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP GetStringLength(REFGUID guidKey, UINT32* pcchLength)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->GetStringLength(guidKey, pcchLength);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP GetString(REFGUID guidKey, LPWSTR pwszValue, UINT32 cchBufSize, UINT32* pcchLength)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->GetString(guidKey, pwszValue, cchBufSize, pcchLength);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP GetAllocatedString(REFGUID guidKey, LPWSTR* ppwszValue, UINT32* pcchLength)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->GetAllocatedString(guidKey, ppwszValue, pcchLength);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP GetBlobSize(REFGUID guidKey, UINT32* pcbBlobSize)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->GetBlobSize(guidKey, pcbBlobSize);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP GetBlob(REFGUID guidKey, UINT8* pBuf, UINT32 cbBufSize, UINT32* pcbBlobSize)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->GetBlob(guidKey, pBuf, cbBufSize, pcbBlobSize);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP GetAllocatedBlob(REFGUID guidKey, UINT8** ppBuf, UINT32* pcbSize)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->GetAllocatedBlob(guidKey, ppBuf, pcbSize);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP GetUnknown(REFGUID guidKey, REFIID riid, LPVOID* ppv)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->GetUnknown(guidKey, riid, ppv);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP SetItem(REFGUID guidKey, REFPROPVARIANT Value)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->SetItem(guidKey, Value);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP DeleteItem(REFGUID guidKey)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->DeleteItem(guidKey);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP DeleteAllItems()
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->DeleteAllItems();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP SetUINT32(REFGUID guidKey, UINT32 unValue)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->SetUINT32(guidKey, unValue);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP SetUINT64(REFGUID guidKey,UINT64 unValue)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->SetUINT64(guidKey, unValue);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP SetDouble(REFGUID guidKey, double fValue)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->SetDouble(guidKey, fValue);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP SetGUID(REFGUID guidKey, REFGUID guidValue)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->SetGUID(guidKey, guidValue);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP SetString(REFGUID guidKey, LPCWSTR wszValue)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->SetString(guidKey, wszValue);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP SetBlob(REFGUID guidKey, const UINT8* pBuf, UINT32 cbBufSize)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->SetBlob(guidKey, pBuf, cbBufSize);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP SetUnknown(REFGUID guidKey, IUnknown* pUnknown)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->SetUnknown(guidKey, pUnknown);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP LockStore()
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->LockStore();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP UnlockStore()
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->UnlockStore();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP GetCount(UINT32* pcItems)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->GetCount(pcItems);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP GetItemByIndex(UINT32 unIndex, GUID* pguidKey, PROPVARIANT* pValue)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->GetItemByIndex(unIndex, pguidKey, pValue);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
STDMETHODIMP CopyAllItems(IMFAttributes* pDest)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return _spAttributes->CopyAllItems(pDest);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Helper functions
|
|
|
|
|
|
|
|
|
|
HRESULT SerializeToStream(DWORD dwOptions, IStream* pStm)
|
|
|
|
|
// dwOptions: Flags from MF_ATTRIBUTE_SERIALIZE_OPTIONS
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return MFSerializeAttributesToStream(_spAttributes.Get(), dwOptions, pStm);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT DeserializeFromStream(DWORD dwOptions, IStream* pStm)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return MFDeserializeAttributesFromStream(_spAttributes.Get(), dwOptions, pStm);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// SerializeToBlob: Stores the attributes in a byte array.
|
|
|
|
|
//
|
|
|
|
|
// ppBuf: Receives a pointer to the byte array.
|
|
|
|
|
// pcbSize: Receives the size of the byte array.
|
|
|
|
|
//
|
|
|
|
|
// The caller must free the array using CoTaskMemFree.
|
|
|
|
|
HRESULT SerializeToBlob(UINT8 **ppBuffer, UINT *pcbSize)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
|
|
|
|
|
if (ppBuffer == NULL)
|
|
|
|
|
{
|
|
|
|
|
return E_POINTER;
|
|
|
|
|
}
|
|
|
|
|
if (pcbSize == NULL)
|
|
|
|
|
{
|
|
|
|
|
return E_POINTER;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
UINT32 cbSize = 0;
|
|
|
|
|
BYTE *pBuffer = NULL;
|
|
|
|
|
|
|
|
|
|
CHECK_HR(hr = MFGetAttributesAsBlobSize(_spAttributes.Get(), &cbSize));
|
|
|
|
|
|
|
|
|
|
pBuffer = (BYTE*)CoTaskMemAlloc(cbSize);
|
|
|
|
|
if (pBuffer == NULL)
|
|
|
|
|
{
|
|
|
|
|
CHECK_HR(hr = E_OUTOFMEMORY);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
CHECK_HR(hr = MFGetAttributesAsBlob(_spAttributes.Get(), pBuffer, cbSize));
|
|
|
|
|
|
|
|
|
|
*ppBuffer = pBuffer;
|
|
|
|
|
*pcbSize = cbSize;
|
|
|
|
|
|
|
|
|
|
done:
|
|
|
|
|
if (FAILED(hr))
|
|
|
|
|
{
|
|
|
|
|
*ppBuffer = NULL;
|
|
|
|
|
*pcbSize = 0;
|
|
|
|
|
CoTaskMemFree(pBuffer);
|
|
|
|
|
}
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT DeserializeFromBlob(const UINT8* pBuffer, UINT cbSize)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return MFInitAttributesFromBlob(_spAttributes.Get(), pBuffer, cbSize);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT GetRatio(REFGUID guidKey, UINT32* pnNumerator, UINT32* punDenominator)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return MFGetAttributeRatio(_spAttributes.Get(), guidKey, pnNumerator, punDenominator);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT SetRatio(REFGUID guidKey, UINT32 unNumerator, UINT32 unDenominator)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return MFSetAttributeRatio(_spAttributes.Get(), guidKey, unNumerator, unDenominator);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Gets an attribute whose value represents the size of something (eg a video frame).
|
|
|
|
|
HRESULT GetSize(REFGUID guidKey, UINT32* punWidth, UINT32* punHeight)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return MFGetAttributeSize(_spAttributes.Get(), guidKey, punWidth, punHeight);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Sets an attribute whose value represents the size of something (eg a video frame).
|
|
|
|
|
HRESULT SetSize(REFGUID guidKey, UINT32 unWidth, UINT32 unHeight)
|
|
|
|
|
{
|
|
|
|
|
assert(_spAttributes);
|
|
|
|
|
return MFSetAttributeSize (_spAttributes.Get(), guidKey, unWidth, unHeight);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
protected:
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFAttributes> _spAttributes;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
};
|
|
|
|
|
|
|
|
|
|
class StreamSink :
|
|
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
public Microsoft::WRL::RuntimeClass<
|
|
|
|
|
Microsoft::WRL::RuntimeClassFlags< Microsoft::WRL::RuntimeClassType::ClassicCom>,
|
|
|
|
|
IMFStreamSink,
|
|
|
|
|
IMFMediaEventGenerator,
|
|
|
|
|
IMFMediaTypeHandler,
|
|
|
|
|
CBaseAttributes<> >
|
|
|
|
|
#else
|
|
|
|
|
public IMFStreamSink,
|
|
|
|
|
public IMFMediaTypeHandler,
|
|
|
|
|
public CBaseAttributes<>,
|
|
|
|
|
public ICustomStreamSink
|
|
|
|
|
#endif
|
|
|
|
|
{
|
|
|
|
|
public:
|
|
|
|
|
// IUnknown methods
|
|
|
|
|
STDMETHOD(QueryInterface)(REFIID riid, _Outptr_result_nullonfailure_ void **ppv)
|
|
|
|
|
{
|
|
|
|
|
if (ppv == nullptr) {
|
|
|
|
|
return E_POINTER;
|
|
|
|
|
}
|
|
|
|
|
(*ppv) = nullptr;
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
if (riid == IID_IMarshal) {
|
|
|
|
|
return MarshalQI(riid, ppv);
|
|
|
|
|
} else {
|
|
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
hr = RuntimeClassT::QueryInterface(riid, ppv);
|
|
|
|
|
#else
|
|
|
|
|
if (riid == IID_IUnknown || riid == IID_IMFStreamSink) {
|
|
|
|
|
*ppv = static_cast<IMFStreamSink*>(this);
|
|
|
|
|
AddRef();
|
|
|
|
|
} else if (riid == IID_IMFMediaEventGenerator) {
|
|
|
|
|
*ppv = static_cast<IMFMediaEventGenerator*>(this);
|
|
|
|
|
AddRef();
|
|
|
|
|
} else if (riid == IID_IMFMediaTypeHandler) {
|
|
|
|
|
*ppv = static_cast<IMFMediaTypeHandler*>(this);
|
|
|
|
|
AddRef();
|
|
|
|
|
} else if (riid == IID_IMFAttributes) {
|
|
|
|
|
*ppv = static_cast<IMFAttributes*>(this);
|
|
|
|
|
AddRef();
|
|
|
|
|
} else if (riid == IID_ICustomStreamSink) {
|
|
|
|
|
*ppv = static_cast<ICustomStreamSink*>(this);
|
|
|
|
|
AddRef();
|
|
|
|
|
} else
|
|
|
|
|
hr = E_NOINTERFACE;
|
|
|
|
|
#endif
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
STDMETHOD(RuntimeClassInitialize)() { return S_OK; }
|
|
|
|
|
#else
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
ULONG STDMETHODCALLTYPE AddRef()
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
{
|
|
|
|
|
return InterlockedIncrement(&m_cRef);
|
|
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
ULONG STDMETHODCALLTYPE Release()
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
{
|
|
|
|
|
ULONG cRef = InterlockedDecrement(&m_cRef);
|
|
|
|
|
if (cRef == 0)
|
|
|
|
|
{
|
|
|
|
|
delete this;
|
|
|
|
|
}
|
|
|
|
|
return cRef;
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
HRESULT MarshalQI(REFIID riid, LPVOID* ppv)
|
|
|
|
|
{
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
if (m_spFTM == nullptr) {
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
if (m_spFTM == nullptr) {
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
hr = CoCreateFreeThreadedMarshaler((IMFStreamSink*)this, &m_spFTM);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
if (m_spFTM == nullptr) {
|
|
|
|
|
hr = E_UNEXPECTED;
|
|
|
|
|
}
|
|
|
|
|
else {
|
|
|
|
|
hr = m_spFTM.Get()->QueryInterface(riid, ppv);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
enum State
|
|
|
|
|
{
|
|
|
|
|
State_TypeNotSet = 0, // No media type is set
|
|
|
|
|
State_Ready, // Media type is set, Start has never been called.
|
|
|
|
|
State_Started,
|
|
|
|
|
State_Stopped,
|
|
|
|
|
State_Paused,
|
|
|
|
|
State_Count // Number of states
|
|
|
|
|
};
|
|
|
|
|
StreamSink() : m_IsShutdown(false),
|
|
|
|
|
m_StartTime(0), m_fGetStartTimeFromSample(false), m_fWaitingForFirstSample(false),
|
|
|
|
|
m_state(State_TypeNotSet), m_pParent(nullptr),
|
|
|
|
|
m_imageWidthInPixels(0), m_imageHeightInPixels(0) {
|
|
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
m_token.value = 0;
|
|
|
|
|
#else
|
|
|
|
|
m_bConnected = false;
|
|
|
|
|
#endif
|
|
|
|
|
InitializeCriticalSectionEx(&m_critSec, 3000, 0);
|
|
|
|
|
ZeroMemory(&m_guiCurrentSubtype, sizeof(m_guiCurrentSubtype));
|
|
|
|
|
CBaseAttributes::Initialize(0U);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::StreamSink\n");
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
|
|
|
|
virtual ~StreamSink() {
|
|
|
|
|
DeleteCriticalSection(&m_critSec);
|
|
|
|
|
assert(m_IsShutdown);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::~StreamSink\n");
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT Initialize()
|
|
|
|
|
{
|
|
|
|
|
HRESULT hr;
|
|
|
|
|
// Create the event queue helper.
|
|
|
|
|
hr = MFCreateEventQueue(&m_spEventQueue);
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFMediaSink> pMedSink;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = CBaseAttributes<>::GetUnknown(MF_STREAMSINK_MEDIASINKINTERFACE, __uuidof(IMFMediaSink), (LPVOID*)pMedSink.GetAddressOf());
|
|
|
|
|
assert(pMedSink.Get() != NULL);
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = pMedSink.Get()->QueryInterface(IID_PPV_ARGS(&m_pParent));
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT CheckShutdown() const
|
|
|
|
|
{
|
|
|
|
|
if (m_IsShutdown)
|
|
|
|
|
{
|
|
|
|
|
return MF_E_SHUTDOWN;
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
return S_OK;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
// Called when the presentation clock starts.
|
|
|
|
|
HRESULT Start(MFTIME start)
|
|
|
|
|
{
|
|
|
|
|
HRESULT hr = S_OK;
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
EnterCriticalSection(&m_critSec);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
if (m_state != State_TypeNotSet) {
|
|
|
|
|
if (start != PRESENTATION_CURRENT_POSITION)
|
|
|
|
|
{
|
|
|
|
|
m_StartTime = start; // Cache the start time.
|
|
|
|
|
m_fGetStartTimeFromSample = false;
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
m_fGetStartTimeFromSample = true;
|
|
|
|
|
}
|
|
|
|
|
m_state = State_Started;
|
|
|
|
|
GUID guiMajorType;
|
|
|
|
|
m_fWaitingForFirstSample = SUCCEEDED(m_spCurrentType->GetMajorType(&guiMajorType)) && (guiMajorType == MFMediaType_Video);
|
|
|
|
|
hr = QueueEvent(MEStreamSinkStarted, GUID_NULL, hr, NULL);
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = QueueEvent(MEStreamSinkRequestSample, GUID_NULL, hr, NULL);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
else hr = MF_E_NOT_INITIALIZED;
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Called when the presentation clock pauses.
|
|
|
|
|
HRESULT Pause()
|
|
|
|
|
{
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
|
|
|
|
|
if (m_state != State_Stopped && m_state != State_TypeNotSet) {
|
|
|
|
|
m_state = State_Paused;
|
|
|
|
|
hr = QueueEvent(MEStreamSinkPaused, GUID_NULL, hr, NULL);
|
|
|
|
|
} else if (hr == State_TypeNotSet)
|
|
|
|
|
hr = MF_E_NOT_INITIALIZED;
|
|
|
|
|
else
|
|
|
|
|
hr = MF_E_INVALIDREQUEST;
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
// Called when the presentation clock restarts.
|
|
|
|
|
HRESULT Restart()
|
|
|
|
|
{
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
|
|
|
|
|
if (m_state == State_Paused) {
|
|
|
|
|
m_state = State_Started;
|
|
|
|
|
hr = QueueEvent(MEStreamSinkStarted, GUID_NULL, hr, NULL);
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = QueueEvent(MEStreamSinkRequestSample, GUID_NULL, hr, NULL);
|
|
|
|
|
}
|
|
|
|
|
} else if (hr == State_TypeNotSet)
|
|
|
|
|
hr = MF_E_NOT_INITIALIZED;
|
|
|
|
|
else
|
|
|
|
|
hr = MF_E_INVALIDREQUEST;
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
// Called when the presentation clock stops.
|
|
|
|
|
HRESULT Stop()
|
|
|
|
|
{
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
if (m_state != State_TypeNotSet) {
|
|
|
|
|
m_state = State_Stopped;
|
|
|
|
|
hr = QueueEvent(MEStreamSinkStopped, GUID_NULL, hr, NULL);
|
|
|
|
|
}
|
|
|
|
|
else hr = MF_E_NOT_INITIALIZED;
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Shuts down the stream sink.
|
|
|
|
|
HRESULT Shutdown()
|
|
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFSampleGrabberSinkCallback> pSampleCallback;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
assert(!m_IsShutdown);
|
|
|
|
|
hr = m_pParent->GetUnknown(MF_MEDIASINK_SAMPLEGRABBERCALLBACK, IID_IMFSampleGrabberSinkCallback, (LPVOID*)pSampleCallback.GetAddressOf());
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = pSampleCallback->OnShutdown();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (m_spEventQueue) {
|
|
|
|
|
hr = m_spEventQueue->Shutdown();
|
|
|
|
|
}
|
|
|
|
|
if (m_pParent)
|
|
|
|
|
m_pParent->Release();
|
|
|
|
|
m_spCurrentType.Reset();
|
|
|
|
|
m_IsShutdown = TRUE;
|
|
|
|
|
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
//IMFStreamSink
|
|
|
|
|
HRESULT STDMETHODCALLTYPE GetMediaSink(
|
|
|
|
|
/* [out] */ __RPC__deref_out_opt IMFMediaSink **ppMediaSink) {
|
|
|
|
|
if (ppMediaSink == NULL)
|
|
|
|
|
{
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFMediaSink> pMedSink;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = CBaseAttributes<>::GetUnknown(MF_STREAMSINK_MEDIASINKINTERFACE, __uuidof(IMFMediaSink), (LPVOID*)pMedSink.GetAddressOf());
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
*ppMediaSink = pMedSink.Detach();
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::GetMediaSink: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE GetIdentifier(
|
|
|
|
|
/* [out] */ __RPC__out DWORD *pdwIdentifier) {
|
|
|
|
|
if (pdwIdentifier == NULL)
|
|
|
|
|
{
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = GetUINT32(MF_STREAMSINK_ID, (UINT32*)pdwIdentifier);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::GetIdentifier: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE GetMediaTypeHandler(
|
|
|
|
|
/* [out] */ __RPC__deref_out_opt IMFMediaTypeHandler **ppHandler) {
|
|
|
|
|
if (ppHandler == NULL)
|
|
|
|
|
{
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
// This stream object acts as its own type handler, so we QI ourselves.
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = QueryInterface(IID_IMFMediaTypeHandler, (void**)ppHandler);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::GetMediaTypeHandler: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE ProcessSample(IMFSample *pSample) {
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFMediaBuffer> pInput;
|
|
|
|
|
_ComPtr<IMFSampleGrabberSinkCallback> pSampleCallback;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
BYTE *pSrc = NULL; // Source buffer.
|
|
|
|
|
// Stride if the buffer does not support IMF2DBuffer
|
|
|
|
|
LONGLONG hnsTime = 0;
|
|
|
|
|
LONGLONG hnsDuration = 0;
|
|
|
|
|
DWORD cbMaxLength;
|
|
|
|
|
DWORD cbCurrentLength = 0;
|
|
|
|
|
GUID guidMajorType;
|
|
|
|
|
if (pSample == NULL)
|
|
|
|
|
{
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
if (m_state != State_Started && m_state != State_Paused) {
|
|
|
|
|
if (m_state == State_TypeNotSet)
|
|
|
|
|
hr = MF_E_NOT_INITIALIZED;
|
|
|
|
|
else
|
|
|
|
|
hr = MF_E_INVALIDREQUEST;
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = CheckShutdown();
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = pSample->ConvertToContiguousBuffer(&pInput);
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = pSample->GetSampleTime(&hnsTime);
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = pSample->GetSampleDuration(&hnsDuration);
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = GetMajorType(&guidMajorType);
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = m_pParent->GetUnknown(MF_MEDIASINK_SAMPLEGRABBERCALLBACK, IID_IMFSampleGrabberSinkCallback, (LPVOID*)pSampleCallback.GetAddressOf());
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = pInput->Lock(&pSrc, &cbMaxLength, &cbCurrentLength);
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = pSampleCallback->OnProcessSample(guidMajorType, 0, hnsTime, hnsDuration, pSrc, cbCurrentLength);
|
|
|
|
|
pInput->Unlock();
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = QueueEvent(MEStreamSinkRequestSample, GUID_NULL, S_OK, NULL);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE PlaceMarker(
|
|
|
|
|
/* [in] */ MFSTREAMSINK_MARKER_TYPE eMarkerType,
|
|
|
|
|
/* [in] */ __RPC__in const PROPVARIANT * /*pvarMarkerValue*/,
|
|
|
|
|
/* [in] */ __RPC__in const PROPVARIANT * /*pvarContextValue*/) {
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
eMarkerType;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
if (m_state == State_TypeNotSet)
|
|
|
|
|
hr = MF_E_NOT_INITIALIZED;
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
//at shutdown will receive MFSTREAMSINK_MARKER_ENDOFSEGMENT
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = QueueEvent(MEStreamSinkRequestSample, GUID_NULL, S_OK, NULL);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::PlaceMarker: HRESULT=%i %s\n", hr, StreamSinkMarkerTypeMap.at(eMarkerType).c_str());
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE Flush(void) {
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::Flush: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
//IMFMediaEventGenerator
|
|
|
|
|
HRESULT STDMETHODCALLTYPE GetEvent(
|
|
|
|
|
DWORD dwFlags, IMFMediaEvent **ppEvent) {
|
|
|
|
|
// NOTE:
|
|
|
|
|
// GetEvent can block indefinitely, so we don't hold the lock.
|
|
|
|
|
// This requires some juggling with the event queue pointer.
|
|
|
|
|
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFMediaEventQueue> pQueue;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
|
|
|
|
|
{
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
// Check shutdown
|
|
|
|
|
hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
// Get the pointer to the event queue.
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
pQueue = m_spEventQueue.Get();
|
|
|
|
|
}
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Now get the event.
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = pQueue->GetEvent(dwFlags, ppEvent);
|
|
|
|
|
}
|
|
|
|
|
MediaEventType meType = MEUnknown;
|
|
|
|
|
if (SUCCEEDED(hr) && SUCCEEDED((*ppEvent)->GetType(&meType)) && meType == MEStreamSinkStopped) {
|
|
|
|
|
}
|
|
|
|
|
HRESULT hrStatus = S_OK;
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = (*ppEvent)->GetStatus(&hrStatus);
|
|
|
|
|
if (SUCCEEDED(hr))
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::GetEvent: HRESULT=%i %s\n", hrStatus, MediaEventTypeMap.at(meType).c_str());
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
else
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::GetEvent: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE BeginGetEvent(
|
|
|
|
|
IMFAsyncCallback *pCallback, IUnknown *punkState) {
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = m_spEventQueue->BeginGetEvent(pCallback, punkState);
|
|
|
|
|
}
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::BeginGetEvent: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE EndGetEvent(
|
|
|
|
|
IMFAsyncResult *pResult, IMFMediaEvent **ppEvent) {
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = m_spEventQueue->EndGetEvent(pResult, ppEvent);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
MediaEventType meType = MEUnknown;
|
|
|
|
|
if (SUCCEEDED(hr) && SUCCEEDED((*ppEvent)->GetType(&meType)) && meType == MEStreamSinkStopped) {
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
|
|
|
|
HRESULT hrStatus = S_OK;
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = (*ppEvent)->GetStatus(&hrStatus);
|
|
|
|
|
if (SUCCEEDED(hr))
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::EndGetEvent: HRESULT=%i %s\n", hrStatus, MediaEventTypeMap.at(meType).c_str());
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
else
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::EndGetEvent: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE QueueEvent(
|
|
|
|
|
MediaEventType met, REFGUID guidExtendedType,
|
|
|
|
|
HRESULT hrStatus, const PROPVARIANT *pvValue) {
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = m_spEventQueue->QueueEventParamVar(met, guidExtendedType, hrStatus, pvValue);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::QueueEvent: HRESULT=%i %s\n", hrStatus, MediaEventTypeMap.at(met).c_str());
|
|
|
|
|
DebugPrintOut(L"StreamSink::QueueEvent: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/// IMFMediaTypeHandler methods
|
|
|
|
|
|
|
|
|
|
// Check if a media type is supported.
|
|
|
|
|
STDMETHODIMP IsMediaTypeSupported(
|
|
|
|
|
/* [in] */ IMFMediaType *pMediaType,
|
|
|
|
|
/* [out] */ IMFMediaType **ppMediaType)
|
|
|
|
|
{
|
|
|
|
|
if (pMediaType == nullptr)
|
|
|
|
|
{
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
GUID majorType = GUID_NULL;
|
|
|
|
|
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = pMediaType->GetGUID(MF_MT_MAJOR_TYPE, &majorType);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// First make sure it's video or audio type.
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
if (majorType != MFMediaType_Video && majorType != MFMediaType_Audio)
|
|
|
|
|
{
|
|
|
|
|
hr = MF_E_INVALIDTYPE;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr) && m_spCurrentType != nullptr)
|
|
|
|
|
{
|
|
|
|
|
GUID guiNewSubtype;
|
|
|
|
|
if (FAILED(pMediaType->GetGUID(MF_MT_SUBTYPE, &guiNewSubtype)) ||
|
|
|
|
|
guiNewSubtype != m_guiCurrentSubtype)
|
|
|
|
|
{
|
|
|
|
|
hr = MF_E_INVALIDTYPE;
|
|
|
|
|
}
|
|
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
// We don't return any "close match" types.
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
if (ppMediaType)
|
|
|
|
|
{
|
|
|
|
|
*ppMediaType = nullptr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (ppMediaType && SUCCEEDED(hr)) {
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFMediaType> pType;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = MFCreateMediaType(ppMediaType);
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = m_pParent->GetUnknown(MF_MEDIASINK_PREFERREDTYPE, __uuidof(IMFMediaType), (LPVOID*)&pType);
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = pType->LockStore();
|
|
|
|
|
}
|
|
|
|
|
bool bLocked = false;
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
bLocked = true;
|
|
|
|
|
UINT32 uiCount;
|
|
|
|
|
UINT32 uiTotal;
|
|
|
|
|
hr = pType->GetCount(&uiTotal);
|
|
|
|
|
for (uiCount = 0; SUCCEEDED(hr) && uiCount < uiTotal; uiCount++) {
|
|
|
|
|
GUID guid;
|
|
|
|
|
PROPVARIANT propval;
|
|
|
|
|
hr = pType->GetItemByIndex(uiCount, &guid, &propval);
|
|
|
|
|
if (SUCCEEDED(hr) && (guid == MF_MT_FRAME_SIZE || guid == MF_MT_MAJOR_TYPE || guid == MF_MT_PIXEL_ASPECT_RATIO ||
|
|
|
|
|
guid == MF_MT_ALL_SAMPLES_INDEPENDENT || guid == MF_MT_INTERLACE_MODE || guid == MF_MT_SUBTYPE)) {
|
|
|
|
|
hr = (*ppMediaType)->SetItem(guid, propval);
|
|
|
|
|
PropVariantClear(&propval);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
if (bLocked) {
|
|
|
|
|
hr = pType->UnlockStore();
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::IsMediaTypeSupported: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// Return the number of preferred media types.
|
|
|
|
|
STDMETHODIMP GetMediaTypeCount(DWORD *pdwTypeCount)
|
|
|
|
|
{
|
|
|
|
|
if (pdwTypeCount == nullptr)
|
|
|
|
|
{
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
// We've got only one media type
|
|
|
|
|
*pdwTypeCount = 1;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::GetMediaTypeCount: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// Return a preferred media type by index.
|
|
|
|
|
STDMETHODIMP GetMediaTypeByIndex(
|
|
|
|
|
/* [in] */ DWORD dwIndex,
|
|
|
|
|
/* [out] */ IMFMediaType **ppType)
|
|
|
|
|
{
|
|
|
|
|
if (ppType == NULL) {
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (dwIndex > 0)
|
|
|
|
|
{
|
|
|
|
|
hr = MF_E_NO_MORE_TYPES;
|
|
|
|
|
} else {
|
|
|
|
|
//return preferred type based on media capture library 6 elements preferred preview type
|
|
|
|
|
//hr = m_spCurrentType.CopyTo(ppType);
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFMediaType> pType;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = MFCreateMediaType(ppType);
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = m_pParent->GetUnknown(MF_MEDIASINK_PREFERREDTYPE, __uuidof(IMFMediaType), (LPVOID*)&pType);
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = pType->LockStore();
|
|
|
|
|
}
|
|
|
|
|
bool bLocked = false;
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
bLocked = true;
|
|
|
|
|
UINT32 uiCount;
|
|
|
|
|
UINT32 uiTotal;
|
|
|
|
|
hr = pType->GetCount(&uiTotal);
|
|
|
|
|
for (uiCount = 0; SUCCEEDED(hr) && uiCount < uiTotal; uiCount++) {
|
|
|
|
|
GUID guid;
|
|
|
|
|
PROPVARIANT propval;
|
|
|
|
|
hr = pType->GetItemByIndex(uiCount, &guid, &propval);
|
|
|
|
|
if (SUCCEEDED(hr) && (guid == MF_MT_FRAME_SIZE || guid == MF_MT_MAJOR_TYPE || guid == MF_MT_PIXEL_ASPECT_RATIO ||
|
|
|
|
|
guid == MF_MT_ALL_SAMPLES_INDEPENDENT || guid == MF_MT_INTERLACE_MODE || guid == MF_MT_SUBTYPE)) {
|
|
|
|
|
hr = (*ppType)->SetItem(guid, propval);
|
|
|
|
|
PropVariantClear(&propval);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
if (bLocked) {
|
|
|
|
|
hr = pType->UnlockStore();
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::GetMediaTypeByIndex: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// Set the current media type.
|
|
|
|
|
STDMETHODIMP SetCurrentMediaType(IMFMediaType *pMediaType)
|
|
|
|
|
{
|
|
|
|
|
if (pMediaType == NULL) {
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
if (m_state != State_TypeNotSet && m_state != State_Ready)
|
|
|
|
|
hr = MF_E_INVALIDREQUEST;
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
// We don't allow format changes after streaming starts.
|
|
|
|
|
|
|
|
|
|
// We set media type already
|
|
|
|
|
if (m_state >= State_Ready)
|
|
|
|
|
{
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = IsMediaTypeSupported(pMediaType, NULL);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = MFCreateMediaType(m_spCurrentType.ReleaseAndGetAddressOf());
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = pMediaType->CopyAllItems(m_spCurrentType.Get());
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = m_spCurrentType->GetGUID(MF_MT_SUBTYPE, &m_guiCurrentSubtype);
|
|
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
GUID guid;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
if (SUCCEEDED(hr)) {
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
hr = m_spCurrentType->GetMajorType(&guid);
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr) && guid == MFMediaType_Video) {
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = MFGetAttributeSize(m_spCurrentType.Get(), MF_MT_FRAME_SIZE, &m_imageWidthInPixels, &m_imageHeightInPixels);
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
m_state = State_Ready;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::SetCurrentMediaType: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Return the current media type, if any.
|
|
|
|
|
STDMETHODIMP GetCurrentMediaType(IMFMediaType **ppMediaType)
|
|
|
|
|
{
|
|
|
|
|
if (ppMediaType == NULL) {
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
if (m_spCurrentType == nullptr) {
|
|
|
|
|
hr = MF_E_NOT_INITIALIZED;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = m_spCurrentType.CopyTo(ppMediaType);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::GetCurrentMediaType: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// Return the major type GUID.
|
|
|
|
|
STDMETHODIMP GetMajorType(GUID *pguidMajorType)
|
|
|
|
|
{
|
|
|
|
|
HRESULT hr;
|
|
|
|
|
if (pguidMajorType == nullptr) {
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFMediaType> pType;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = m_pParent->GetUnknown(MF_MEDIASINK_PREFERREDTYPE, __uuidof(IMFMediaType), (LPVOID*)&pType);
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = pType->GetMajorType(pguidMajorType);
|
|
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"StreamSink::GetMajorType: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
private:
|
|
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
EventRegistrationToken m_token;
|
|
|
|
|
#else
|
|
|
|
|
bool m_bConnected;
|
|
|
|
|
#endif
|
|
|
|
|
|
|
|
|
|
bool m_IsShutdown; // Flag to indicate if Shutdown() method was called.
|
|
|
|
|
CRITICAL_SECTION m_critSec;
|
|
|
|
|
#ifndef HAVE_WINRT
|
|
|
|
|
long m_cRef;
|
|
|
|
|
#endif
|
|
|
|
|
IMFAttributes* m_pParent;
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFMediaType> m_spCurrentType;
|
|
|
|
|
_ComPtr<IMFMediaEventQueue> m_spEventQueue; // Event queue
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IUnknown> m_spFTM;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
State m_state;
|
|
|
|
|
bool m_fGetStartTimeFromSample;
|
|
|
|
|
bool m_fWaitingForFirstSample;
|
|
|
|
|
MFTIME m_StartTime; // Presentation time when the clock started.
|
|
|
|
|
GUID m_guiCurrentSubtype;
|
|
|
|
|
UINT32 m_imageWidthInPixels;
|
|
|
|
|
UINT32 m_imageHeightInPixels;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
// Notes:
|
|
|
|
|
//
|
|
|
|
|
// The List class template implements a simple double-linked list.
|
|
|
|
|
// It uses STL's copy semantics.
|
|
|
|
|
|
|
|
|
|
// There are two versions of the Clear() method:
|
|
|
|
|
// Clear(void) clears the list w/out cleaning up the object.
|
|
|
|
|
// Clear(FN fn) takes a functor object that releases the objects, if they need cleanup.
|
|
|
|
|
|
|
|
|
|
// The List class supports enumeration. Example of usage:
|
|
|
|
|
//
|
|
|
|
|
// List<T>::POSIITON pos = list.GetFrontPosition();
|
|
|
|
|
// while (pos != list.GetEndPosition())
|
|
|
|
|
// {
|
|
|
|
|
// T item;
|
|
|
|
|
// hr = list.GetItemPos(&item);
|
|
|
|
|
// pos = list.Next(pos);
|
|
|
|
|
// }
|
|
|
|
|
|
|
|
|
|
// The ComPtrList class template derives from List<> and implements a list of COM pointers.
|
|
|
|
|
|
|
|
|
|
template <class T>
|
|
|
|
|
struct NoOp
|
|
|
|
|
{
|
|
|
|
|
void operator()(T& /*t*/)
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
template <class T>
|
|
|
|
|
class List
|
|
|
|
|
{
|
|
|
|
|
protected:
|
|
|
|
|
|
|
|
|
|
// Nodes in the linked list
|
|
|
|
|
struct Node
|
|
|
|
|
{
|
|
|
|
|
Node *prev;
|
|
|
|
|
Node *next;
|
|
|
|
|
T item;
|
|
|
|
|
|
|
|
|
|
Node() : prev(nullptr), next(nullptr)
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
Node(T item) : prev(nullptr), next(nullptr)
|
|
|
|
|
{
|
|
|
|
|
this->item = item;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
T Item() const { return item; }
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
public:
|
|
|
|
|
|
|
|
|
|
// Object for enumerating the list.
|
|
|
|
|
class POSITION
|
|
|
|
|
{
|
|
|
|
|
friend class List<T>;
|
|
|
|
|
|
|
|
|
|
public:
|
|
|
|
|
POSITION() : pNode(nullptr)
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
bool operator==(const POSITION &p) const
|
|
|
|
|
{
|
|
|
|
|
return pNode == p.pNode;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
bool operator!=(const POSITION &p) const
|
|
|
|
|
{
|
|
|
|
|
return pNode != p.pNode;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
private:
|
|
|
|
|
const Node *pNode;
|
|
|
|
|
|
|
|
|
|
POSITION(Node *p) : pNode(p)
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
protected:
|
|
|
|
|
Node m_anchor; // Anchor node for the linked list.
|
|
|
|
|
DWORD m_count; // Number of items in the list.
|
|
|
|
|
|
|
|
|
|
Node* Front() const
|
|
|
|
|
{
|
|
|
|
|
return m_anchor.next;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
Node* Back() const
|
|
|
|
|
{
|
|
|
|
|
return m_anchor.prev;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
virtual HRESULT InsertAfter(T item, Node *pBefore)
|
|
|
|
|
{
|
|
|
|
|
if (pBefore == nullptr)
|
|
|
|
|
{
|
|
|
|
|
return E_POINTER;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
Node *pNode = new Node(item);
|
|
|
|
|
if (pNode == nullptr)
|
|
|
|
|
{
|
|
|
|
|
return E_OUTOFMEMORY;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
Node *pAfter = pBefore->next;
|
|
|
|
|
|
|
|
|
|
pBefore->next = pNode;
|
|
|
|
|
pAfter->prev = pNode;
|
|
|
|
|
|
|
|
|
|
pNode->prev = pBefore;
|
|
|
|
|
pNode->next = pAfter;
|
|
|
|
|
|
|
|
|
|
m_count++;
|
|
|
|
|
|
|
|
|
|
return S_OK;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
virtual HRESULT GetItem(const Node *pNode, T* ppItem)
|
|
|
|
|
{
|
|
|
|
|
if (pNode == nullptr || ppItem == nullptr)
|
|
|
|
|
{
|
|
|
|
|
return E_POINTER;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
*ppItem = pNode->item;
|
|
|
|
|
return S_OK;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// RemoveItem:
|
|
|
|
|
// Removes a node and optionally returns the item.
|
|
|
|
|
// ppItem can be nullptr.
|
|
|
|
|
virtual HRESULT RemoveItem(Node *pNode, T *ppItem)
|
|
|
|
|
{
|
|
|
|
|
if (pNode == nullptr)
|
|
|
|
|
{
|
|
|
|
|
return E_POINTER;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
assert(pNode != &m_anchor); // We should never try to remove the anchor node.
|
|
|
|
|
if (pNode == &m_anchor)
|
|
|
|
|
{
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
T item;
|
|
|
|
|
|
|
|
|
|
// The next node's previous is this node's previous.
|
|
|
|
|
pNode->next->prev = pNode->prev;
|
|
|
|
|
|
|
|
|
|
// The previous node's next is this node's next.
|
|
|
|
|
pNode->prev->next = pNode->next;
|
|
|
|
|
|
|
|
|
|
item = pNode->item;
|
|
|
|
|
delete pNode;
|
|
|
|
|
|
|
|
|
|
m_count--;
|
|
|
|
|
|
|
|
|
|
if (ppItem)
|
|
|
|
|
{
|
|
|
|
|
*ppItem = item;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return S_OK;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
public:
|
|
|
|
|
|
|
|
|
|
List()
|
|
|
|
|
{
|
|
|
|
|
m_anchor.next = &m_anchor;
|
|
|
|
|
m_anchor.prev = &m_anchor;
|
|
|
|
|
|
|
|
|
|
m_count = 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
virtual ~List()
|
|
|
|
|
{
|
|
|
|
|
Clear();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Insertion functions
|
|
|
|
|
HRESULT InsertBack(T item)
|
|
|
|
|
{
|
|
|
|
|
return InsertAfter(item, m_anchor.prev);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
HRESULT InsertFront(T item)
|
|
|
|
|
{
|
|
|
|
|
return InsertAfter(item, &m_anchor);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT InsertPos(POSITION pos, T item)
|
|
|
|
|
{
|
|
|
|
|
if (pos.pNode == nullptr)
|
|
|
|
|
{
|
|
|
|
|
return InsertBack(item);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return InsertAfter(item, pos.pNode->prev);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// RemoveBack: Removes the tail of the list and returns the value.
|
|
|
|
|
// ppItem can be nullptr if you don't want the item back. (But the method does not release the item.)
|
|
|
|
|
HRESULT RemoveBack(T *ppItem)
|
|
|
|
|
{
|
|
|
|
|
if (IsEmpty())
|
|
|
|
|
{
|
|
|
|
|
return E_FAIL;
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
return RemoveItem(Back(), ppItem);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// RemoveFront: Removes the head of the list and returns the value.
|
|
|
|
|
// ppItem can be nullptr if you don't want the item back. (But the method does not release the item.)
|
|
|
|
|
HRESULT RemoveFront(T *ppItem)
|
|
|
|
|
{
|
|
|
|
|
if (IsEmpty())
|
|
|
|
|
{
|
|
|
|
|
return E_FAIL;
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
return RemoveItem(Front(), ppItem);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// GetBack: Gets the tail item.
|
|
|
|
|
HRESULT GetBack(T *ppItem)
|
|
|
|
|
{
|
|
|
|
|
if (IsEmpty())
|
|
|
|
|
{
|
|
|
|
|
return E_FAIL;
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
return GetItem(Back(), ppItem);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// GetFront: Gets the front item.
|
|
|
|
|
HRESULT GetFront(T *ppItem)
|
|
|
|
|
{
|
|
|
|
|
if (IsEmpty())
|
|
|
|
|
{
|
|
|
|
|
return E_FAIL;
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
return GetItem(Front(), ppItem);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// GetCount: Returns the number of items in the list.
|
|
|
|
|
DWORD GetCount() const { return m_count; }
|
|
|
|
|
|
|
|
|
|
bool IsEmpty() const
|
|
|
|
|
{
|
|
|
|
|
return (GetCount() == 0);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Clear: Takes a functor object whose operator()
|
|
|
|
|
// frees the object on the list.
|
|
|
|
|
template <class FN>
|
|
|
|
|
void Clear(FN& clear_fn)
|
|
|
|
|
{
|
|
|
|
|
Node *n = m_anchor.next;
|
|
|
|
|
|
|
|
|
|
// Delete the nodes
|
|
|
|
|
while (n != &m_anchor)
|
|
|
|
|
{
|
|
|
|
|
clear_fn(n->item);
|
|
|
|
|
|
|
|
|
|
Node *tmp = n->next;
|
|
|
|
|
delete n;
|
|
|
|
|
n = tmp;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Reset the anchor to point at itself
|
|
|
|
|
m_anchor.next = &m_anchor;
|
|
|
|
|
m_anchor.prev = &m_anchor;
|
|
|
|
|
|
|
|
|
|
m_count = 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Clear: Clears the list. (Does not delete or release the list items.)
|
|
|
|
|
virtual void Clear()
|
|
|
|
|
{
|
|
|
|
|
NoOp<T> clearOp;
|
|
|
|
|
Clear<>(clearOp);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// Enumerator functions
|
|
|
|
|
|
|
|
|
|
POSITION FrontPosition()
|
|
|
|
|
{
|
|
|
|
|
if (IsEmpty())
|
|
|
|
|
{
|
|
|
|
|
return POSITION(nullptr);
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
return POSITION(Front());
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
POSITION EndPosition() const
|
|
|
|
|
{
|
|
|
|
|
return POSITION();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT GetItemPos(POSITION pos, T *ppItem)
|
|
|
|
|
{
|
|
|
|
|
if (pos.pNode)
|
|
|
|
|
{
|
|
|
|
|
return GetItem(pos.pNode, ppItem);
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
return E_FAIL;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
POSITION Next(const POSITION pos)
|
|
|
|
|
{
|
|
|
|
|
if (pos.pNode && (pos.pNode->next != &m_anchor))
|
|
|
|
|
{
|
|
|
|
|
return POSITION(pos.pNode->next);
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
return POSITION(nullptr);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Remove an item at a position.
|
|
|
|
|
// The item is returns in ppItem, unless ppItem is nullptr.
|
|
|
|
|
// NOTE: This method invalidates the POSITION object.
|
|
|
|
|
HRESULT Remove(POSITION& pos, T *ppItem)
|
|
|
|
|
{
|
|
|
|
|
if (pos.pNode)
|
|
|
|
|
{
|
|
|
|
|
// Remove const-ness temporarily...
|
|
|
|
|
Node *pNode = const_cast<Node*>(pos.pNode);
|
|
|
|
|
|
|
|
|
|
pos = POSITION();
|
|
|
|
|
|
|
|
|
|
return RemoveItem(pNode, ppItem);
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// Typical functors for Clear method.
|
|
|
|
|
|
|
|
|
|
// ComAutoRelease: Releases COM pointers.
|
|
|
|
|
// MemDelete: Deletes pointers to new'd memory.
|
|
|
|
|
|
|
|
|
|
class ComAutoRelease
|
|
|
|
|
{
|
|
|
|
|
public:
|
|
|
|
|
void operator()(IUnknown *p)
|
|
|
|
|
{
|
|
|
|
|
if (p)
|
|
|
|
|
{
|
|
|
|
|
p->Release();
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
class MemDelete
|
|
|
|
|
{
|
|
|
|
|
public:
|
|
|
|
|
void operator()(void *p)
|
|
|
|
|
{
|
|
|
|
|
if (p)
|
|
|
|
|
{
|
|
|
|
|
delete p;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
// ComPtrList class
|
|
|
|
|
// Derived class that makes it safer to store COM pointers in the List<> class.
|
|
|
|
|
// It automatically AddRef's the pointers that are inserted onto the list
|
|
|
|
|
// (unless the insertion method fails).
|
|
|
|
|
//
|
|
|
|
|
// T must be a COM interface type.
|
|
|
|
|
// example: ComPtrList<IUnknown>
|
|
|
|
|
//
|
|
|
|
|
// NULLABLE: If true, client can insert nullptr pointers. This means GetItem can
|
|
|
|
|
// succeed but return a nullptr pointer. By default, the list does not allow nullptr
|
|
|
|
|
// pointers.
|
|
|
|
|
|
|
|
|
|
template <class T, bool NULLABLE = FALSE>
|
|
|
|
|
class ComPtrList : public List<T*>
|
|
|
|
|
{
|
|
|
|
|
public:
|
|
|
|
|
|
|
|
|
|
typedef T* Ptr;
|
|
|
|
|
|
|
|
|
|
void Clear()
|
|
|
|
|
{
|
|
|
|
|
ComAutoRelease car;
|
|
|
|
|
List<Ptr>::Clear(car);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
~ComPtrList()
|
|
|
|
|
{
|
|
|
|
|
Clear();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
protected:
|
|
|
|
|
HRESULT InsertAfter(Ptr item, Node *pBefore)
|
|
|
|
|
{
|
|
|
|
|
// Do not allow nullptr item pointers unless NULLABLE is true.
|
|
|
|
|
if (item == nullptr && !NULLABLE)
|
|
|
|
|
{
|
|
|
|
|
return E_POINTER;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (item)
|
|
|
|
|
{
|
|
|
|
|
item->AddRef();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT hr = List<Ptr>::InsertAfter(item, pBefore);
|
|
|
|
|
if (FAILED(hr) && item != nullptr)
|
|
|
|
|
{
|
|
|
|
|
item->Release();
|
|
|
|
|
}
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT GetItem(const Node *pNode, Ptr* ppItem)
|
|
|
|
|
{
|
|
|
|
|
Ptr pItem = nullptr;
|
|
|
|
|
|
|
|
|
|
// The base class gives us the pointer without AddRef'ing it.
|
|
|
|
|
// If we return the pointer to the caller, we must AddRef().
|
|
|
|
|
HRESULT hr = List<Ptr>::GetItem(pNode, &pItem);
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
assert(pItem || NULLABLE);
|
|
|
|
|
if (pItem)
|
|
|
|
|
{
|
|
|
|
|
*ppItem = pItem;
|
|
|
|
|
(*ppItem)->AddRef();
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT RemoveItem(Node *pNode, Ptr *ppItem)
|
|
|
|
|
{
|
|
|
|
|
// ppItem can be nullptr, but we need to get the
|
|
|
|
|
// item so that we can release it.
|
|
|
|
|
|
|
|
|
|
// If ppItem is not nullptr, we will AddRef it on the way out.
|
|
|
|
|
|
|
|
|
|
Ptr pItem = nullptr;
|
|
|
|
|
|
|
|
|
|
HRESULT hr = List<Ptr>::RemoveItem(pNode, &pItem);
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
assert(pItem || NULLABLE);
|
|
|
|
|
if (ppItem && pItem)
|
|
|
|
|
{
|
|
|
|
|
*ppItem = pItem;
|
|
|
|
|
(*ppItem)->AddRef();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (pItem)
|
|
|
|
|
{
|
|
|
|
|
pItem->Release();
|
|
|
|
|
pItem = nullptr;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
/* Be sure to declare webcam device capability in manifest
|
|
|
|
|
For better media capture support, add the following snippet with correct module name to the project manifest
|
|
|
|
|
(videoio needs DLL activation class factoryentry points):
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
<Extensions>
|
|
|
|
|
<Extension Category="windows.activatableClass.inProcessServer">
|
|
|
|
|
<InProcessServer>
|
|
|
|
|
<Path>modulename</Path>
|
|
|
|
|
<ActivatableClass ActivatableClassId="cv.MediaSink" ThreadingModel="both" />
|
|
|
|
|
</InProcessServer>
|
|
|
|
|
</Extension>
|
|
|
|
|
</Extensions>*/
|
|
|
|
|
|
|
|
|
|
extern const __declspec(selectany) WCHAR RuntimeClass_CV_MediaSink[] = L"cv.MediaSink";
|
|
|
|
|
|
|
|
|
|
class MediaSink :
|
|
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
public Microsoft::WRL::RuntimeClass<
|
|
|
|
|
Microsoft::WRL::RuntimeClassFlags< Microsoft::WRL::RuntimeClassType::WinRtClassicComMix >,
|
|
|
|
|
Microsoft::WRL::Implements<ABI::Windows::Media::IMediaExtension>,
|
|
|
|
|
IMFMediaSink,
|
|
|
|
|
IMFClockStateSink,
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
Microsoft::WRL::FtmBase,
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
CBaseAttributes<>>
|
|
|
|
|
#else
|
|
|
|
|
public IMFMediaSink, public IMFClockStateSink, public CBaseAttributes<>
|
|
|
|
|
#endif
|
|
|
|
|
{
|
|
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
InspectableClass(RuntimeClass_CV_MediaSink, BaseTrust)
|
|
|
|
|
public:
|
|
|
|
|
#else
|
|
|
|
|
public:
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
ULONG STDMETHODCALLTYPE AddRef()
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
{
|
|
|
|
|
return InterlockedIncrement(&m_cRef);
|
|
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
ULONG STDMETHODCALLTYPE Release()
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
{
|
|
|
|
|
ULONG cRef = InterlockedDecrement(&m_cRef);
|
|
|
|
|
if (cRef == 0)
|
|
|
|
|
{
|
|
|
|
|
delete this;
|
|
|
|
|
}
|
|
|
|
|
return cRef;
|
|
|
|
|
}
|
|
|
|
|
STDMETHOD(QueryInterface)(REFIID riid, _Outptr_result_nullonfailure_ void **ppv)
|
|
|
|
|
{
|
|
|
|
|
if (ppv == nullptr) {
|
|
|
|
|
return E_POINTER;
|
|
|
|
|
}
|
|
|
|
|
(*ppv) = nullptr;
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
if (riid == IID_IUnknown ||
|
|
|
|
|
riid == IID_IMFMediaSink) {
|
|
|
|
|
(*ppv) = static_cast<IMFMediaSink*>(this);
|
|
|
|
|
AddRef();
|
|
|
|
|
} else if (riid == IID_IMFClockStateSink) {
|
|
|
|
|
(*ppv) = static_cast<IMFClockStateSink*>(this);
|
|
|
|
|
AddRef();
|
|
|
|
|
} else if (riid == IID_IMFAttributes) {
|
|
|
|
|
(*ppv) = static_cast<IMFAttributes*>(this);
|
|
|
|
|
AddRef();
|
|
|
|
|
} else {
|
|
|
|
|
hr = E_NOINTERFACE;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
MediaSink() : m_IsShutdown(false), m_llStartTime(0) {
|
|
|
|
|
CBaseAttributes<>::Initialize(0U);
|
|
|
|
|
InitializeCriticalSectionEx(&m_critSec, 3000, 0);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::MediaSink\n");
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
|
|
|
|
|
|
|
|
|
virtual ~MediaSink() {
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::~MediaSink\n");
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
DeleteCriticalSection(&m_critSec);
|
|
|
|
|
assert(m_IsShutdown);
|
|
|
|
|
}
|
|
|
|
|
HRESULT CheckShutdown() const
|
|
|
|
|
{
|
|
|
|
|
if (m_IsShutdown)
|
|
|
|
|
{
|
|
|
|
|
return MF_E_SHUTDOWN;
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
return S_OK;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
STDMETHODIMP SetProperties(ABI::Windows::Foundation::Collections::IPropertySet *pConfiguration)
|
|
|
|
|
{
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
if (pConfiguration) {
|
|
|
|
|
Microsoft::WRL::ComPtr<IInspectable> spInsp;
|
|
|
|
|
Microsoft::WRL::ComPtr<ABI::Windows::Foundation::Collections::IMap<HSTRING, IInspectable *>> spSetting;
|
|
|
|
|
Microsoft::WRL::ComPtr<ABI::Windows::Foundation::IPropertyValue> spPropVal;
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
Microsoft::WRL::ComPtr<ABI::Windows::Media::MediaProperties::IMediaEncodingProperties> pMedEncProps;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
UINT32 uiType = ABI::Windows::Media::Capture::MediaStreamType_VideoPreview;
|
|
|
|
|
|
|
|
|
|
hr = pConfiguration->QueryInterface(IID_PPV_ARGS(&spSetting));
|
|
|
|
|
if (FAILED(hr)) {
|
|
|
|
|
hr = E_FAIL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
hr = spSetting->Lookup(Microsoft::WRL::Wrappers::HStringReference(MF_PROP_SAMPLEGRABBERCALLBACK).Get(), spInsp.ReleaseAndGetAddressOf());
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
if (FAILED(hr)) {
|
|
|
|
|
hr = E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = SetUnknown(MF_MEDIASINK_SAMPLEGRABBERCALLBACK, spInsp.Get());
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
hr = spSetting->Lookup(Microsoft::WRL::Wrappers::HStringReference(MF_PROP_VIDTYPE).Get(), spInsp.ReleaseAndGetAddressOf());
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
if (FAILED(hr)) {
|
|
|
|
|
hr = E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
if (SUCCEEDED(hr = spInsp.As(&spPropVal))) {
|
|
|
|
|
hr = spPropVal->GetUInt32(&uiType);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
hr = spSetting->Lookup(Microsoft::WRL::Wrappers::HStringReference(MF_PROP_VIDENCPROPS).Get(), spInsp.ReleaseAndGetAddressOf());
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
if (FAILED(hr)) {
|
|
|
|
|
hr = E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = spInsp.As(&pMedEncProps);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = SetMediaStreamProperties((ABI::Windows::Media::Capture::MediaStreamType)uiType, pMedEncProps.Get());
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
static DWORD GetStreamId(ABI::Windows::Media::Capture::MediaStreamType mediaStreamType)
|
|
|
|
|
{
|
|
|
|
|
return 3 - mediaStreamType;
|
|
|
|
|
}
|
|
|
|
|
static HRESULT AddAttribute(_In_ GUID guidKey, _In_ ABI::Windows::Foundation::IPropertyValue *pValue, _In_ IMFAttributes* pAttr)
|
|
|
|
|
{
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
PROPVARIANT var;
|
|
|
|
|
ABI::Windows::Foundation::PropertyType type;
|
|
|
|
|
hr = pValue->get_Type(&type);
|
|
|
|
|
ZeroMemory(&var, sizeof(var));
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
switch (type)
|
|
|
|
|
{
|
|
|
|
|
case ABI::Windows::Foundation::PropertyType_UInt8Array:
|
|
|
|
|
{
|
|
|
|
|
UINT32 cbBlob;
|
|
|
|
|
BYTE *pbBlog = nullptr;
|
|
|
|
|
hr = pValue->GetUInt8Array(&cbBlob, &pbBlog);
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
if (pbBlog == nullptr)
|
|
|
|
|
{
|
|
|
|
|
hr = E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
hr = pAttr->SetBlob(guidKey, pbBlog, cbBlob);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
CoTaskMemFree(pbBlog);
|
|
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
case ABI::Windows::Foundation::PropertyType_Double:
|
|
|
|
|
{
|
|
|
|
|
DOUBLE value;
|
|
|
|
|
hr = pValue->GetDouble(&value);
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = pAttr->SetDouble(guidKey, value);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
case ABI::Windows::Foundation::PropertyType_Guid:
|
|
|
|
|
{
|
|
|
|
|
GUID value;
|
|
|
|
|
hr = pValue->GetGuid(&value);
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = pAttr->SetGUID(guidKey, value);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
case ABI::Windows::Foundation::PropertyType_String:
|
|
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
Microsoft::WRL::Wrappers::HString value;
|
|
|
|
|
hr = pValue->GetString(value.GetAddressOf());
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
UINT32 len = 0;
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
LPCWSTR szValue = WindowsGetStringRawBuffer(value.Get(), &len);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = pAttr->SetString(guidKey, szValue);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
case ABI::Windows::Foundation::PropertyType_UInt32:
|
|
|
|
|
{
|
|
|
|
|
UINT32 value;
|
|
|
|
|
hr = pValue->GetUInt32(&value);
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
pAttr->SetUINT32(guidKey, value);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
case ABI::Windows::Foundation::PropertyType_UInt64:
|
|
|
|
|
{
|
|
|
|
|
UINT64 value;
|
|
|
|
|
hr = pValue->GetUInt64(&value);
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = pAttr->SetUINT64(guidKey, value);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
case ABI::Windows::Foundation::PropertyType_Inspectable:
|
|
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
Microsoft::WRL::ComPtr<IInspectable> value;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = TYPE_E_TYPEMISMATCH;
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
pAttr->SetUnknown(guidKey, value.Get());
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
break;
|
|
|
|
|
|
|
|
|
|
// ignore unknown values
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
static HRESULT ConvertPropertiesToMediaType(_In_ ABI::Windows::Media::MediaProperties::IMediaEncodingProperties *pMEP, _Outptr_ IMFMediaType **ppMT)
|
|
|
|
|
{
|
|
|
|
|
HRESULT hr = S_OK;
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFMediaType> spMT;
|
|
|
|
|
Microsoft::WRL::ComPtr<ABI::Windows::Foundation::Collections::IMap<GUID, IInspectable*>> spMap;
|
|
|
|
|
Microsoft::WRL::ComPtr<ABI::Windows::Foundation::Collections::IIterable<ABI::Windows::Foundation::Collections::IKeyValuePair<GUID, IInspectable*>*>> spIterable;
|
|
|
|
|
Microsoft::WRL::ComPtr<ABI::Windows::Foundation::Collections::IIterator<ABI::Windows::Foundation::Collections::IKeyValuePair<GUID, IInspectable*>*>> spIterator;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
|
|
|
|
|
if (pMEP == nullptr || ppMT == nullptr)
|
|
|
|
|
{
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
*ppMT = nullptr;
|
|
|
|
|
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
hr = pMEP->get_Properties(spMap.GetAddressOf());
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = spMap.As(&spIterable);
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = spIterable->First(&spIterator);
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
MFCreateMediaType(spMT.ReleaseAndGetAddressOf());
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
boolean hasCurrent = false;
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = spIterator->get_HasCurrent(&hasCurrent);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
while (hasCurrent)
|
|
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
Microsoft::WRL::ComPtr<ABI::Windows::Foundation::Collections::IKeyValuePair<GUID, IInspectable*> > spKeyValuePair;
|
|
|
|
|
Microsoft::WRL::ComPtr<IInspectable> spValue;
|
|
|
|
|
Microsoft::WRL::ComPtr<ABI::Windows::Foundation::IPropertyValue> spPropValue;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
GUID guidKey;
|
|
|
|
|
|
|
|
|
|
hr = spIterator->get_Current(&spKeyValuePair);
|
|
|
|
|
if (FAILED(hr))
|
|
|
|
|
{
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
hr = spKeyValuePair->get_Key(&guidKey);
|
|
|
|
|
if (FAILED(hr))
|
|
|
|
|
{
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
hr = spKeyValuePair->get_Value(&spValue);
|
|
|
|
|
if (FAILED(hr))
|
|
|
|
|
{
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
hr = spValue.As(&spPropValue);
|
|
|
|
|
if (FAILED(hr))
|
|
|
|
|
{
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
hr = AddAttribute(guidKey, spPropValue.Get(), spMT.Get());
|
|
|
|
|
if (FAILED(hr))
|
|
|
|
|
{
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
hr = spIterator->MoveNext(&hasCurrent);
|
|
|
|
|
if (FAILED(hr))
|
|
|
|
|
{
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
Microsoft::WRL::ComPtr<IInspectable> spValue;
|
|
|
|
|
Microsoft::WRL::ComPtr<ABI::Windows::Foundation::IPropertyValue> spPropValue;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
GUID guiMajorType;
|
|
|
|
|
|
|
|
|
|
hr = spMap->Lookup(MF_MT_MAJOR_TYPE, spValue.GetAddressOf());
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = spValue.As(&spPropValue);
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = spPropValue->GetGuid(&guiMajorType);
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
if (guiMajorType != MFMediaType_Video && guiMajorType != MFMediaType_Audio)
|
|
|
|
|
{
|
|
|
|
|
hr = E_UNEXPECTED;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
*ppMT = spMT.Detach();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
//this should be passed through SetProperties!
|
|
|
|
|
HRESULT SetMediaStreamProperties(ABI::Windows::Media::Capture::MediaStreamType MediaStreamType,
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
_In_opt_ ABI::Windows::Media::MediaProperties::IMediaEncodingProperties *mediaEncodingProperties)
|
|
|
|
|
{
|
|
|
|
|
HRESULT hr = S_OK;
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFMediaType> spMediaType;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
|
|
|
|
|
if (MediaStreamType != ABI::Windows::Media::Capture::MediaStreamType_VideoPreview &&
|
|
|
|
|
MediaStreamType != ABI::Windows::Media::Capture::MediaStreamType_VideoRecord &&
|
|
|
|
|
MediaStreamType != ABI::Windows::Media::Capture::MediaStreamType_Audio)
|
|
|
|
|
{
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
RemoveStreamSink(GetStreamId(MediaStreamType));
|
|
|
|
|
|
|
|
|
|
if (mediaEncodingProperties != nullptr)
|
|
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFStreamSink> spStreamSink;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = ConvertPropertiesToMediaType(mediaEncodingProperties, &spMediaType);
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = AddStreamSink(GetStreamId(MediaStreamType), nullptr, spStreamSink.GetAddressOf());
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = SetUnknown(MF_MEDIASINK_PREFERREDTYPE, spMediaType.Detach());
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
//IMFMediaSink
|
|
|
|
|
HRESULT STDMETHODCALLTYPE GetCharacteristics(
|
|
|
|
|
/* [out] */ __RPC__out DWORD *pdwCharacteristics) {
|
|
|
|
|
HRESULT hr;
|
|
|
|
|
if (pdwCharacteristics == NULL) return E_INVALIDARG;
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
if (SUCCEEDED(hr = CheckShutdown())) {
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
//if had an activation object for the sink, shut down would be managed and MF_STREAM_SINK_SUPPORTS_ROTATION appears to be setable to TRUE
|
|
|
|
|
*pdwCharacteristics = MEDIASINK_FIXED_STREAMS;// | MEDIASINK_REQUIRE_REFERENCE_MEDIATYPE;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::GetCharacteristics: HRESULT=%i\n", hr);
|
|
|
|
|
return hr;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE AddStreamSink(
|
|
|
|
|
DWORD dwStreamSinkIdentifier, IMFMediaType * /*pMediaType*/, IMFStreamSink **ppStreamSink) {
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFStreamSink> spMFStream;
|
|
|
|
|
_ComPtr<ICustomStreamSink> pStream;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = GetStreamSinkById(dwStreamSinkIdentifier, &spMFStream);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = MF_E_STREAMSINK_EXISTS;
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
hr = S_OK;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
pStream = Microsoft::WRL::Make<StreamSink>();
|
|
|
|
|
if (pStream == nullptr) {
|
|
|
|
|
hr = E_OUTOFMEMORY;
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = pStream.As<IMFStreamSink>(&spMFStream);
|
|
|
|
|
#else
|
|
|
|
|
StreamSink* pSink = new StreamSink();
|
|
|
|
|
if (pSink) {
|
|
|
|
|
hr = pSink->QueryInterface(IID_IMFStreamSink, (void**)spMFStream.GetAddressOf());
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = spMFStream.As(&pStream);
|
|
|
|
|
}
|
|
|
|
|
if (FAILED(hr)) delete pSink;
|
|
|
|
|
}
|
|
|
|
|
#endif
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Initialize the stream.
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFAttributes> pAttr;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = pStream.As(&pAttr);
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = pAttr->SetUINT32(MF_STREAMSINK_ID, dwStreamSinkIdentifier);
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = pAttr->SetUnknown(MF_STREAMSINK_MEDIASINKINTERFACE, (IMFMediaSink*)this);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
hr = pStream->Initialize();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
ComPtrList<IMFStreamSink>::POSITION pos = m_streams.FrontPosition();
|
|
|
|
|
ComPtrList<IMFStreamSink>::POSITION posEnd = m_streams.EndPosition();
|
|
|
|
|
|
|
|
|
|
// Insert in proper position
|
|
|
|
|
for (; pos != posEnd; pos = m_streams.Next(pos))
|
|
|
|
|
{
|
|
|
|
|
DWORD dwCurrId;
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFStreamSink> spCurr;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = m_streams.GetItemPos(pos, &spCurr);
|
|
|
|
|
if (FAILED(hr))
|
|
|
|
|
{
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
hr = spCurr->GetIdentifier(&dwCurrId);
|
|
|
|
|
if (FAILED(hr))
|
|
|
|
|
{
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (dwCurrId > dwStreamSinkIdentifier)
|
|
|
|
|
{
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = m_streams.InsertPos(pos, spMFStream.Get());
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
*ppStreamSink = spMFStream.Detach();
|
|
|
|
|
}
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::AddStreamSink: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE RemoveStreamSink(DWORD dwStreamSinkIdentifier) {
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
ComPtrList<IMFStreamSink>::POSITION pos = m_streams.FrontPosition();
|
|
|
|
|
ComPtrList<IMFStreamSink>::POSITION endPos = m_streams.EndPosition();
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFStreamSink> spStream;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
for (; pos != endPos; pos = m_streams.Next(pos))
|
|
|
|
|
{
|
|
|
|
|
hr = m_streams.GetItemPos(pos, &spStream);
|
|
|
|
|
DWORD dwId;
|
|
|
|
|
|
|
|
|
|
if (FAILED(hr))
|
|
|
|
|
{
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
hr = spStream->GetIdentifier(&dwId);
|
|
|
|
|
if (FAILED(hr) || dwId == dwStreamSinkIdentifier)
|
|
|
|
|
{
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (pos == endPos)
|
|
|
|
|
{
|
|
|
|
|
hr = MF_E_INVALIDSTREAMNUMBER;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
hr = m_streams.Remove(pos, nullptr);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<ICustomStreamSink> spCustomSink;
|
|
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
spCustomSink = static_cast<StreamSink*>(spStream.Get());
|
|
|
|
|
hr = S_OK;
|
|
|
|
|
#else
|
|
|
|
|
hr = spStream.As(&spCustomSink);
|
|
|
|
|
#endif
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = spCustomSink->Shutdown();
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::RemoveStreamSink: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE GetStreamSinkCount(DWORD *pStreamSinkCount) {
|
|
|
|
|
if (pStreamSinkCount == NULL)
|
|
|
|
|
{
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
*pStreamSinkCount = m_streams.GetCount();
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::GetStreamSinkCount: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE GetStreamSinkByIndex(
|
|
|
|
|
DWORD dwIndex, IMFStreamSink **ppStreamSink) {
|
|
|
|
|
if (ppStreamSink == NULL)
|
|
|
|
|
{
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFStreamSink> spStream;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
DWORD cStreams = m_streams.GetCount();
|
|
|
|
|
|
|
|
|
|
if (dwIndex >= cStreams)
|
|
|
|
|
{
|
|
|
|
|
return MF_E_INVALIDINDEX;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
ComPtrList<IMFStreamSink>::POSITION pos = m_streams.FrontPosition();
|
|
|
|
|
ComPtrList<IMFStreamSink>::POSITION endPos = m_streams.EndPosition();
|
|
|
|
|
DWORD dwCurrent = 0;
|
|
|
|
|
|
|
|
|
|
for (; pos != endPos && dwCurrent < dwIndex; pos = m_streams.Next(pos), ++dwCurrent)
|
|
|
|
|
{
|
|
|
|
|
// Just move to proper position
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (pos == endPos)
|
|
|
|
|
{
|
|
|
|
|
hr = MF_E_UNEXPECTED;
|
|
|
|
|
}
|
|
|
|
|
else
|
|
|
|
|
{
|
|
|
|
|
hr = m_streams.GetItemPos(pos, &spStream);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
*ppStreamSink = spStream.Detach();
|
|
|
|
|
}
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::GetStreamSinkByIndex: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE GetStreamSinkById(
|
|
|
|
|
DWORD dwStreamSinkIdentifier, IMFStreamSink **ppStreamSink) {
|
|
|
|
|
if (ppStreamSink == NULL)
|
|
|
|
|
{
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFStreamSink> spResult;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
ComPtrList<IMFStreamSink>::POSITION pos = m_streams.FrontPosition();
|
|
|
|
|
ComPtrList<IMFStreamSink>::POSITION endPos = m_streams.EndPosition();
|
|
|
|
|
|
|
|
|
|
for (; pos != endPos; pos = m_streams.Next(pos))
|
|
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFStreamSink> spStream;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = m_streams.GetItemPos(pos, &spStream);
|
|
|
|
|
DWORD dwId;
|
|
|
|
|
|
|
|
|
|
if (FAILED(hr))
|
|
|
|
|
{
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
hr = spStream->GetIdentifier(&dwId);
|
|
|
|
|
if (FAILED(hr))
|
|
|
|
|
{
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
else if (dwId == dwStreamSinkIdentifier)
|
|
|
|
|
{
|
|
|
|
|
spResult = spStream;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (pos == endPos)
|
|
|
|
|
{
|
|
|
|
|
hr = MF_E_INVALIDSTREAMNUMBER;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
assert(spResult);
|
|
|
|
|
*ppStreamSink = spResult.Detach();
|
|
|
|
|
}
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::GetStreamSinkById: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE SetPresentationClock(
|
|
|
|
|
IMFPresentationClock *pPresentationClock) {
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
// If we already have a clock, remove ourselves from that clock's
|
|
|
|
|
// state notifications.
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
if (m_spClock) {
|
|
|
|
|
hr = m_spClock->RemoveClockStateSink(this);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
// Register ourselves to get state notifications from the new clock.
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
if (pPresentationClock) {
|
|
|
|
|
hr = pPresentationClock->AddClockStateSink(this);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFSampleGrabberSinkCallback> pSampleCallback;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
// Release the pointer to the old clock.
|
|
|
|
|
// Store the pointer to the new clock.
|
|
|
|
|
m_spClock = pPresentationClock;
|
|
|
|
|
hr = GetUnknown(MF_MEDIASINK_SAMPLEGRABBERCALLBACK, IID_IMFSampleGrabberSinkCallback, (LPVOID*)pSampleCallback.GetAddressOf());
|
|
|
|
|
}
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = pSampleCallback->OnSetPresentationClock(pPresentationClock);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::SetPresentationClock: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE GetPresentationClock(
|
|
|
|
|
IMFPresentationClock **ppPresentationClock) {
|
|
|
|
|
if (ppPresentationClock == NULL) {
|
|
|
|
|
return E_INVALIDARG;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
if (m_spClock == NULL) {
|
|
|
|
|
hr = MF_E_NO_CLOCK; // There is no presentation clock.
|
|
|
|
|
} else {
|
|
|
|
|
// Return the pointer to the caller.
|
|
|
|
|
hr = m_spClock.CopyTo(ppPresentationClock);
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::GetPresentationClock: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE Shutdown(void) {
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
|
|
|
|
ForEach(m_streams, ShutdownFunc());
|
|
|
|
|
m_streams.Clear();
|
|
|
|
|
m_spClock.ReleaseAndGetAddressOf();
|
|
|
|
|
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFMediaType> pType;
|
|
|
|
|
hr = CBaseAttributes<>::GetUnknown(MF_MEDIASINK_PREFERREDTYPE, __uuidof(IMFMediaType), (LPVOID*)pType.GetAddressOf());
|
|
|
|
|
if (SUCCEEDED(hr)) {
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = DeleteItem(MF_MEDIASINK_PREFERREDTYPE);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
}
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
m_IsShutdown = true;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::Shutdown: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
class ShutdownFunc
|
|
|
|
|
{
|
|
|
|
|
public:
|
|
|
|
|
HRESULT operator()(IMFStreamSink *pStream) const
|
|
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<ICustomStreamSink> spCustomSink;
|
|
|
|
|
HRESULT hr;
|
|
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
spCustomSink = static_cast<StreamSink*>(pStream);
|
|
|
|
|
#else
|
|
|
|
|
hr = pStream->QueryInterface(IID_PPV_ARGS(spCustomSink.GetAddressOf()));
|
|
|
|
|
if (FAILED(hr)) return hr;
|
|
|
|
|
#endif
|
|
|
|
|
hr = spCustomSink->Shutdown();
|
|
|
|
|
return hr;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
class StartFunc
|
|
|
|
|
{
|
|
|
|
|
public:
|
|
|
|
|
StartFunc(LONGLONG llStartTime)
|
|
|
|
|
: _llStartTime(llStartTime)
|
|
|
|
|
{
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT operator()(IMFStreamSink *pStream) const
|
|
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<ICustomStreamSink> spCustomSink;
|
|
|
|
|
HRESULT hr;
|
|
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
spCustomSink = static_cast<StreamSink*>(pStream);
|
|
|
|
|
#else
|
|
|
|
|
hr = pStream->QueryInterface(IID_PPV_ARGS(spCustomSink.GetAddressOf()));
|
|
|
|
|
if (FAILED(hr)) return hr;
|
|
|
|
|
#endif
|
|
|
|
|
hr = spCustomSink->Start(_llStartTime);
|
|
|
|
|
return hr;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
|
|
|
|
|
|
|
|
|
LONGLONG _llStartTime;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
class StopFunc
|
|
|
|
|
{
|
|
|
|
|
public:
|
|
|
|
|
HRESULT operator()(IMFStreamSink *pStream) const
|
|
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<ICustomStreamSink> spCustomSink;
|
|
|
|
|
HRESULT hr;
|
|
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
spCustomSink = static_cast<StreamSink*>(pStream);
|
|
|
|
|
#else
|
|
|
|
|
hr = pStream->QueryInterface(IID_PPV_ARGS(spCustomSink.GetAddressOf()));
|
|
|
|
|
if (FAILED(hr)) return hr;
|
|
|
|
|
#endif
|
|
|
|
|
hr = spCustomSink->Stop();
|
|
|
|
|
return hr;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
}
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
template <class T, class TFunc>
|
|
|
|
|
HRESULT ForEach(ComPtrList<T> &col, TFunc fn)
|
|
|
|
|
{
|
|
|
|
|
ComPtrList<T>::POSITION pos = col.FrontPosition();
|
|
|
|
|
ComPtrList<T>::POSITION endPos = col.EndPosition();
|
|
|
|
|
HRESULT hr = S_OK;
|
|
|
|
|
|
|
|
|
|
for (; pos != endPos; pos = col.Next(pos))
|
|
|
|
|
{
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<T> spStream;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
|
|
|
|
|
hr = col.GetItemPos(pos, &spStream);
|
|
|
|
|
if (FAILED(hr))
|
|
|
|
|
{
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
hr = fn(spStream.Get());
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
//IMFClockStateSink
|
|
|
|
|
HRESULT STDMETHODCALLTYPE OnClockStart(
|
|
|
|
|
MFTIME hnsSystemTime,
|
|
|
|
|
LONGLONG llClockStartOffset) {
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
// Start each stream.
|
|
|
|
|
m_llStartTime = llClockStartOffset;
|
|
|
|
|
hr = ForEach(m_streams, StartFunc(llClockStartOffset));
|
|
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFSampleGrabberSinkCallback> pSampleCallback;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = GetUnknown(MF_MEDIASINK_SAMPLEGRABBERCALLBACK, IID_IMFSampleGrabberSinkCallback, (LPVOID*)pSampleCallback.GetAddressOf());
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = pSampleCallback->OnClockStart(hnsSystemTime, llClockStartOffset);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::OnClockStart: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE OnClockStop(
|
|
|
|
|
MFTIME hnsSystemTime) {
|
|
|
|
|
EnterCriticalSection(&m_critSec);
|
|
|
|
|
HRESULT hr = CheckShutdown();
|
|
|
|
|
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
{
|
|
|
|
|
// Stop each stream
|
|
|
|
|
hr = ForEach(m_streams, StopFunc());
|
|
|
|
|
}
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFSampleGrabberSinkCallback> pSampleCallback;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = GetUnknown(MF_MEDIASINK_SAMPLEGRABBERCALLBACK, IID_IMFSampleGrabberSinkCallback, (LPVOID*)pSampleCallback.GetAddressOf());
|
|
|
|
|
LeaveCriticalSection(&m_critSec);
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = pSampleCallback->OnClockStop(hnsSystemTime);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::OnClockStop: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE OnClockPause(
|
|
|
|
|
MFTIME hnsSystemTime) {
|
|
|
|
|
HRESULT hr;
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFSampleGrabberSinkCallback> pSampleCallback;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = GetUnknown(MF_MEDIASINK_SAMPLEGRABBERCALLBACK, IID_IMFSampleGrabberSinkCallback, (LPVOID*)pSampleCallback.GetAddressOf());
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = pSampleCallback->OnClockPause(hnsSystemTime);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::OnClockPause: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE OnClockRestart(
|
|
|
|
|
MFTIME hnsSystemTime) {
|
|
|
|
|
HRESULT hr;
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFSampleGrabberSinkCallback> pSampleCallback;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = GetUnknown(MF_MEDIASINK_SAMPLEGRABBERCALLBACK, IID_IMFSampleGrabberSinkCallback, (LPVOID*)pSampleCallback.GetAddressOf());
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = pSampleCallback->OnClockRestart(hnsSystemTime);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::OnClockRestart: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
HRESULT STDMETHODCALLTYPE OnClockSetRate(
|
|
|
|
|
MFTIME hnsSystemTime,
|
|
|
|
|
float flRate) {
|
|
|
|
|
HRESULT hr;
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFSampleGrabberSinkCallback> pSampleCallback;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
hr = GetUnknown(MF_MEDIASINK_SAMPLEGRABBERCALLBACK, IID_IMFSampleGrabberSinkCallback, (LPVOID*)pSampleCallback.GetAddressOf());
|
|
|
|
|
if (SUCCEEDED(hr))
|
|
|
|
|
hr = pSampleCallback->OnClockSetRate(hnsSystemTime, flRate);
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
DebugPrintOut(L"MediaSink::OnClockSetRate: HRESULT=%i\n", hr);
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
return hr;
|
|
|
|
|
}
|
|
|
|
|
private:
|
|
|
|
|
#ifndef HAVE_WINRT
|
|
|
|
|
long m_cRef;
|
|
|
|
|
#endif
|
|
|
|
|
CRITICAL_SECTION m_critSec;
|
|
|
|
|
bool m_IsShutdown;
|
|
|
|
|
ComPtrList<IMFStreamSink> m_streams;
|
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update and rename cap_msmf.h to cap_msmf.hpp
Successful test - samples are grabbed
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Fixed trailing whitespace
VS 2013 support and cleanup consistency plus C++/CX new object fixed
Conflicts:
modules/highgui/src/cap_msmf.cpp
modules/highgui/src/cap_msmf.hpp
modules/highgui/src/ppltasks_winrt.h
Fix merge conflicts
VS 2013 Update 2 library bug fix integrated
a-wi's changed integrated
11 years ago
|
|
|
_ComPtr<IMFPresentationClock> m_spClock;
|
Update cap_msmf.cpp
Add support for WinRT in the MF capture framework by removing the disallowed calls to enumerate devices and create a sample grabber sink and adding framework for the MediaCapture interface and a custom sink which interfaces with the sample grabber callback interface. The change requires discussion for making it completely functional as redundancy is required given that if the source is a video file, the old code pathways must be used. Otherwise all IMFMediaSession, IMFMediaSource, and IMFActivate code must use a MediaCapture code path and all sink code must use the CMediaSink custom sink.
Support for the custom sink is extended to non-WinRT not for compatibility as Windows Vista client is a minimum regardless, but because it offers more flexibility, could be faster and is able to be used as an optionally different code path during sink creation based on a future configuration parameter.
My discussion and proposal to finish this change:
Devices are so easily enumerated through WinRT Windows.Devices namespace that wrapping the calls in a library is quite a chore for little benefit though to get the various modes and formats could still be a worthwhile project. For now conditional compilation to remove videodevices and any offending non-video file related activity in videodevice. In my opinion, this is a different , far less fundamental and important change which can possibly be done as a future project and also much more easily implemented in C++/CX.
ImageGrabber has the IMFSampleGrabberSinkCallback replaced with a base class (SharedSampleGrabber) which also be is base class for ImageGrabberRT. This change is necessary as the custom sink does not require a thread to pump events which is done through MediaCapture already. IMFSampleGrabberSinkCallback is the common element between both models and that piece can be shared. Initializing the new ImageGrabberRT is as simple as passing an already initialized MediaCapture object and any video format/encoding parameters.
The concurrency event is necessary to wait for completion and is the way the underlying, IAsyncAction wrappers in the task library work as well. Native WIN32 event objects would be an option if HAVE_CONCURRENCY is not defined. I could even imagine doing it with sleep/thread yield and InterlockedCompareExchange yet I am not enthusiastic about that approach either. Since there is a specific compiler HAVE_ for concurrency, I do not like pulling it in though I think for WinRT it is safe to say we will always have it available though should probably conditionally compile with the Interlocked option as WIN32 events would require HAVE_WIN32.
It looks like C++/CX cannot be used for the IMediaExtension sink (which should not be a problem) as using COM objects requires WRL and though deriving from IMediaExtension can be done, there is little purpose without COM. Objects from C++/CX can be swapped to interact with objects from native C++ as Inspectable* can reinterpret_cast to the ref object IInspectable^ and vice-versa. A solution to the COM class with C++/CX would be great so we could have dual support. Also without #define for every WRL object in use, the code will get quite muddy given that the */^ would need to be ifdef'd everywhere.
Update cap_msmf.cpp
Fixed bugs and completed the change. I believe the new classes need to be moved to a header file as the file has become to large and more classes need to be added for handling all the asynchronous problems (one wrapping IAsyncAction in a task and another for making a task out of IAsyncAction). Unfortunately, blocking on the UI thread is not an option in WinRT so a synchronous architecture is considered "illegal" by Microsoft's standards even if implementable (C++/CX ppltasks library throws errors if you try it). Worse, either by design or a bug in the MF MediaCapture class with Custom Sinks causes a crash if stop/start previewing without reinitializing (spPreferredPreviewMediaType is fatally nulled). After decompiling Windows.Media.dll, I worked around this in my own projects by using an activate-able custom sink ID which strangely assigns 1 to this pointer allowing it to be reinitialized in what can only be described as a hack by Microsoft. This would add additional overhead to the project to implement especially for static libraries as it requires IDL/DLL exporting followed by manifest declaration. Better to document that it is not supported.
Furthermore, an additional class for IMFAttributes should be implemented to make clean architecture for passing around attributes as opposed to directly calling non-COM interface calls on the objects and making use of SetProperties which would also be a set up for an object that uses the RuntimeClass activation ID.
The remaining changes are not difficult and will be complete soon along with debug tracing messages.
Update cap_msmf.cpp
Create cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update cap_msmf.cpp
Update cap_msmf.h
Update and rename cap_msmf.h to cap_msmf.hpp
Update cap_msmf.cpp
Update CMakeLists.txt
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Successful test - samples are grabbed
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update cap_msmf.hpp
Create ppltasks_winrt.h
Update cap_msmf.hpp
Update cap_msmf.hpp
Update cap_msmf.cpp
Update ppltasks_winrt.h
Library updated and cleaned up with comments, marshaling, exceptions and linker settings
Update ppltasks_winrt.h
Fixed trailing whitespace
11 years ago
|
|
|
LONGLONG m_llStartTime;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
#ifdef HAVE_WINRT
|
|
|
|
|
ActivatableClass(MediaSink);
|
|
|
|
|
#endif
|
