[mythtv] FFmpeg sync

[Janne Grunau]

Hi,

It's FFmpeg sync time. I made patches and my git mirror[1] of mythtv's 
svn repository available to testers in the #mythtv irc channel on 
freenode.

I think the sync is pretty stable and I'm only aware of 
one "regression". The volume of FFmpeg's native AC3 decoder after 
downmixing 5.1 audio to stereo. I've compared the implementation with 
the specs and I'm pretty sure that the implemantion is correct.
Previously we amplified the audio during downmixing but I don't think it 
belongs there. It looks as if the code correct and more importly the 
AC3 can use now a SIMD implemation of the downmixing function.

So I'm writing a software amplifier in mythtv instead. I initially will 
use it only for downmixed AC3 audio but it also be used for extended 
volume control. I hope to have that ready in the next few days.

AC3 passthrough with the native decoder was changed and is afaik 
untested.

The FFmpeg devs have just increased the major libavcodec version and are 
currently dropping deprecated parts. Some of them (software scaler and 
colorspace converted, ...) are used by mythtv.

The most important changes for mythtv in FFmpeg are E-AC3 decoder, H264 
decoding fixes, and native AAC decoder. The AAC decoder misses features 
to make libfaad obsolete though.

http://www.grunau.be/mythtv_ffmpeg_sync_15219_18295.diff.bz2

the patch has to applied with "patch -p1" to a complete trunk checkout 
since it needs modify some plugins too.

Alternatively the branch ffmpeg/latest of my git tree could be used.


I plan to apply soon if no regressions are reported and someone verifies 
that AC3 passthrough still works.


Janne

[1] git clone git://git.jannau.net/mythtv.git, only trunk and 0.21 fixes 
are tracked