r20879: [small fixes]
r20933: libvorbis is not necessary for Vorbis playback, small wording improvement.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@20939 b3059339-0415-0410-9bf9-f77b7e298cf2
Pass video codec extradata unchanged from demux_real, sync vd_realvid to
the new format
Sync mkv demuxer to the changes above (cmsg24 extradata was totally
broken before)
Detect cmsg24 size from extradata (was fixed)
Based on a patch by elupus >> elupus >a< ecce se <<
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@20936 b3059339-0415-0410-9bf9-f77b7e298cf2
Patch by Carl Eugen Hoyos [cehoyos <at> ag or at] with modifications by me.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@20931 b3059339-0415-0410-9bf9-f77b7e298cf2
Having a look at the gcc output before committing is a good idea btw.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@20920 b3059339-0415-0410-9bf9-f77b7e298cf2
The most visible inaccuracy caused by the previous update location was
that the OSD always showed position 0 after seeking with demux_mkv.
Split frame decoding and filtering because with -correct-pts the pts
value that should be displayed for the frame is only known after
decoding but is needed before filtering (during which the OSD is drawn).
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@20918 b3059339-0415-0410-9bf9-f77b7e298cf2
Fixes some problems with playback_speed!=1: sleep limits which were
supposed to be real time were calculated in scaled time, timing
inaccuracies in nosound mode affected next frame by
err*(1-1/playback_speed), auto quality code used scaled time.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@20915 b3059339-0415-0410-9bf9-f77b7e298cf2
If the shown frame was already at exactly the end pos we don't want to
show one more frame after it.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@20911 b3059339-0415-0410-9bf9-f77b7e298cf2
This code was apparently intended to duplicate frames in constant-
framerate output in case of skipped input frames, but the test used
(blit_frame not set at that location) is not correct for that use.
If removing this code turns out to make the brokenness of other code
more apparent then a better fix might be needed.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@20908 b3059339-0415-0410-9bf9-f77b7e298cf2
Larger buffers could make sense with lots of high-bitrate audio
channels.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@20906 b3059339-0415-0410-9bf9-f77b7e298cf2
text (missing styles and edl support) subtitles in mov,
see also bug #611
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@20893 b3059339-0415-0410-9bf9-f77b7e298cf2
This fix removes semitransparent area (less then pixel width) between glyph and
it's outline. Instead, it makes them overlap a little. It usually looks much
better this way.
Complete fix seems impossible with the current output format (single color
alpha bitmaps). The right way is to blend both glyph and outline into one
bitmap so that 2 pixels with 50% transparency produce a fully solid one.
This requires RGBA bitmap output from libass.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@20890 b3059339-0415-0410-9bf9-f77b7e298cf2
