avoid duplicate checks for formats by changeing them (Y8->Y800, IYUV->I420)
Originally committed as revision 4482 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc
x/yInc bugfix (bug was introduced during the global var removial)
more emms & no messing with ARCH_X86 (=workaround against not cleared mmx state somewhere)
sharpening filter (-ssf cs=<int> & -ssf ls=<int>)
Originally committed as revision 4298 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc
-ssf cgb=<float> -> chrominance gaussian blur (nice to improve picture quality)
-ssf lgb=<float> -> luminance gaussian blur (dunno what its good for)
-ssf chs=<int> -> shift chrominance horizontally
-ssf cvs=<int> -> shift chrominance vertically
these will only work if the swscaler is used at all
Originally committed as revision 4295 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc
dynamic memory allocation for the buffers (needed for the filter stuff)
Originally committed as revision 4291 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc
experimental upscaling mode (-sws 3)
general convolution filters support (unfinished)
bugfix for bicubic upscaling
assertion checking if defined MP_DEBUG
checking of the input/output size instead of segfault if its very large
Originally committed as revision 4277 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc
force quantizer (for these lq movies which where reencoded at high bitrates)
(c) 2002
Originally committed as revision 3965 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc
chrominance & luminance filtering is the default now if -npp ...:y isnt specified (lumi only was only used for the deinterlacers and that was wrong anyway)
Originally committed as revision 3818 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc
filterPos was wrong by 0.5 pixels on downscale with -sws 1 and 2 bugfix
Originally committed as revision 3783 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc
see line527@postproc/postprocess.c ;) for the documentation of the format
Originally committed as revision 3772 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc
use new horizontal mmx scaler instead of old x86asm if mmx2 cant be used (FAST_BILINEAR only)
fixed overflow in init function ... using double precission fp now :)
using C scaler for the last 1-2 lines if there is a chance to write over the end of the dst array
Originally committed as revision 3353 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc
-sws command line option (0 -> fast_bilinear, 1->bilinear, 2->bicubic)
Originally committed as revision 3273 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc
green line at bottom bugfix
green lines in yuv2yuv scaler bugfix
Originally committed as revision 3210 to svn://svn.mplayerhq.hu/mplayer/trunk/postproc
Michael Niedermayer
Felix Bünemann
Alban Bedel
Johannes Feigl