[Ffmpeg-devel] Re: [RFC] mpeg2 422 encoding.

Baptiste COUDURIER baptiste.coudurier
Sat Apr 1 16:07:36 CEST 2006


Hi

Michael Niedermayer wrote:
> Hi
> 
> On Fri, Mar 31, 2006 at 08:37:43PM -0500, Rich Felker wrote:
>> On Sat, Apr 01, 2006 at 03:03:37AM +0200, Baptiste COUDURIER wrote:
>>> Baptiste COUDURIER wrote:
>>>> [...]
>>> Ok, a bit more. It works correctly for intra. It produces picture for P
>>> and B frames, though containing artifacts like blue/red blocks.
>>> Bistream is correct.
>>>
>>> Im digging more into the code. It may be related to motion estimation.
>>> Michael, does motion estimation work in 422 chroma format ? If it is,
>>> then Im a bit lost here.
>> AFAIK motion estimation ignores chroma channels unless you override it
>> with a special option.
> 
> exactly
> 
> also motion estimation doesnt matter, set all MVs to arbitrary valid vectors
> and it should still work, i mean quality and bitrate will suck but there
> shouldnt be any blue or red blocks
> 
> here are a few random ideas ...
> try -debug 16384 it will vissually show MB types
> try -vismv ... it will show motion vectors

I get errors(segv,dbl free) when using vismv and debug on 422, I need to
fix.

> try to force vectors or MB types to specific ones

forcing CANDIDATE_MB_INTRA makes P and B frames clean.

> try valgrind

valgrind says everything is ok.

> do P frames with just intra blocks work?

P and B with just intra block works.

> do P frames with just inter blocks with MV 0,0 work?

forcing CANDIDATE_MB_INTER and motion_x, motion_y to 0 produces many
artefacts, blue/red green/purple blocks.

> [...]


-- 
Baptiste COUDURIER                              GnuPG Key Id: 0x5C1ABAAA
SMARTJOG S.A.                                    http://www.smartjog.com
Key fingerprint                 8D77134D20CC9220201FC5DB0AC9325C5C1ABAAA
Phone: +33 1 49966312





More information about the ffmpeg-devel mailing list