frame, there will be only one centroid and it will be impossible to
find three distinct ones. It is better to just avoid shifting if
there are not three different centroids.
Originally committed as revision 14343 to svn://svn.ffmpeg.org/ffmpeg/trunk
that has no corresponding points. Not only it is the
worst possible pick, but also the code was written
without this case in mind.
Originally committed as revision 14341 to svn://svn.ffmpeg.org/ffmpeg/trunk
different codebook centroids ("high utility", "low
utility" and "closest to the low utility one"). This
change avoid the corner case of choosing two times the
same centroid.
Originally committed as revision 14340 to svn://svn.ffmpeg.org/ffmpeg/trunk
Fixes at least:
src19td.IBP.264
CVWP3_TOSHIBA_E.264
cvmp_mot_picaff0_full_B.26l
CVMP_MOT_FRM_L31_B.26l
cvmp_mot_frm0_full_B.26l
CVMP_MOT_FLD_L30_B.26l
cvmp_mot_fld0_full_B.26l
Originally committed as revision 14337 to svn://svn.ffmpeg.org/ffmpeg/trunk
optimization, more interresting would it have been had the author
thought about what value chroma_qp would have for the following MB.
Or failing that, had actually tested the code.
So this reverts this non-functional optimization, and makes the code work.
Fixes at least CAPM3_Sony_D.jsv
Originally committed as revision 14335 to svn://svn.ffmpeg.org/ffmpeg/trunk
Fixes at least:
CABAST3_Sony_E.jsv
CABASTBR3_Sony_A.jsv
CABASTBR3_Sony_B.jsv
Originally committed as revision 14331 to svn://svn.ffmpeg.org/ffmpeg/trunk
This way we use slice_type_nos almost everywhere which means 1 variable less
for gcc to put in a register.
Originally committed as revision 14326 to svn://svn.ffmpeg.org/ffmpeg/trunk
This includes indentation changes, comment reformatting, consistent brace
placement and some prettyprinting.
Originally committed as revision 14318 to svn://svn.ffmpeg.org/ffmpeg/trunk
This includes indentation changes, comment reformatting, consistent brace
placement and some prettyprinting.
Originally committed as revision 14316 to svn://svn.ffmpeg.org/ffmpeg/trunk
Disable filter_mb_fast() as it optimized the incorrect code.
Fixes at least:
BA3_SVA_C.264
CABA3_SVA_B.264
CABACI3_Sony_B.jsv
CAFI1_SVA_C.264
camp_mot_frm0_full.26l
CAWP5_TOSHIBA_E.264
CVFI2_SVA_C.264
CVSE3_Sony_H.jsv
CVWP2_TOSHIBA_E.264
CVWP5_TOSHIBA_E.264
SL1_SVA_B.264
Originally committed as revision 14315 to svn://svn.ffmpeg.org/ffmpeg/trunk
That is, add 16 frames delay, cache trashing and av desync.
fixes at least the following reference bitstreams:
CABA3_Sony_C.jsv
CACQP3_Sony_D.jsv
CAMANL1_TOSHIBA_B.264
CANL3_Sony_C.jsv
CVBS3_Sony_C.jsv
CVMANL1_TOSHIBA_B.264
Originally committed as revision 14308 to svn://svn.ffmpeg.org/ffmpeg/trunk
AVOptions, actually it works even finer as it also gets passed to the
decoder.
Originally committed as revision 14307 to svn://svn.ffmpeg.org/ffmpeg/trunk
same maximum that can be achieved by specifying the value in the bitstream.
Originally committed as revision 14302 to svn://svn.ffmpeg.org/ffmpeg/trunk
Kostya Shishkov
Vitor Sessak
Michael Niedermayer
tripp
Carl Eugen Hoyos
Stefano Sabatini
Diego Biurrun
Erik Hovland
Andreas Öman
Justin Ruggles
Benjamin Larsson