[MPlayer-dev-eng] [Fwd: [Ffmpeg-devel] reproducible problem with mencoder encoding with libavcodec (again, plus new info)]

[Corey Hickey]

Hi,

I've been lurking on here and I saw you were going to be rounding up
user bugreports. I've sent variations of this to both mplayer-users and
ffmpeg-devel. If you want any more details, testing, etc, feel free to
ask.

Thanks,
Corey

-------- Original Message --------
Subject: [Ffmpeg-devel] reproducible problem with mencoder encoding with 
libavcodec (again, plus new info)
Date: Thu, 07 Aug 2003 18:45:06 -0700
From: Corey Hickey <bugfood-ml at fatooh.org>
Reply-To: ffmpeg-devel at lists.sourceforge.net
To:  <ffmpeg-devel at lists.sourceforge.net>

Hello,

A little while back I posted a bugreport to the mplayer-users list:

http://mplayerhq.hu/pipermail/mplayer-users/2003-August/035825.html

I also checked here on ffmpeg-devel, and Michael Niedermayer had a look
and said it looked like an mplayer bug. I never received any feedback on
mplayer-users, but since then Hans Meine refered to a report he had made
earlier:

http://www.mplayerhq.hu/pipermail/mplayer-users/2003-July/034941.html

He reports on what seems to be the same problem, but when using bframes.
In that case the divx2pass.log file gets mostly filled with zeroes
nstead of mencoder crashing.

Since then I went ahead and tracked down the exact CVS commits that seem
to cause this problem. I used "cvs ... -D" with both mplayer and ffmpeg
cvs, and found that it seems to be the changes made to ffmpeg cvs at Fri
Apr 25 00:29:48 2003 UTC, with log message "fixing mixed dr1 + internal
buffers".

Any mplayer build with libavcodec cvs from that point forward seems to
fail to encode the file I posted in my original bugreport. I noticed
that mplayer cvs has a patch to vd_ffmpeg.c made shortly thereafter:

http://www1.mplayerhq.hu/cgi-bin/cvsweb.cgi/main/libmpcodecs/vd_ffmpeg.c.diff?r1=1.87&r2=1.88

...but, that doesn't seem to solve the problem.

I hope this information is helpful, and thanks again.

-Corey