Audio timestamps are passed through by default and when the input
doesnt contain clean timestamps this can lead to non monotonicity
errors. (rounding to a course timebase can cause this too)
Print a warning when the errors in the timestamps are large
Fixes Ticket1167 (regression since timestamps are passed through)
This is a generic workaround that is intended to handle
slightly incorrect input files. It is very possible that some
demuxers contain bugs that lead to wrong timestamps, these demuxers
should of course still be fixed even if this change happens to
hide the issue.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
dr1 seems to work fine with frame size changes but many filters
cant handle it yet. Simply disabling it forces the alternative
non dr1 code path which has been tested more completely and
is known to handle frame size changes in a wider varity of
cases.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
Allows working around issue #605.
Note: as a side effect this fixes that -vsync drop
as far as I could tell would not drop pts/dts values
when duplicating frames or when flushing encoder delay.
Signed-off-by: Reimar Döffinger <Reimar.Doeffinger@gmx.de>
With this we can always know if a timestamp is based on added durations
from an unknown origin or if it is based on a correct timestamp (and possibly
added durations)
This should fix some bugs where this distinction was mixed up.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
commit 13f6917ca9 handles discards automatically,
but the ffm discard options are not fully parsed. Causing the input streams not
to be used, so no stream towards the ffserver after the initial probing.
Signed-off-by: Rick van der Zwet <info@rickvanderzwet.nl>
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
This is somewhat redundant as no decoder should call get_buffer() with such argument.
Found-by: Mateusz "j00ru" Jurczyk and Gynvael Coldwind
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
This probably fixes some of the use of uninitialized issues valgrind shows in fate.
It might also fix other issues.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
input_video_filter and output_video_filter can't be NULL at this point.
If they are, the current code would likely crash anyway (since
filtered_frame would be NULL and sent to do_video_out().