[XviD-devel] gmc problem - who to blame? version 2.0

[Christoph Lampert]

Okay, this time it's 2 bframes and Quarterpel, right? 
Quarterpel+GMC is not tested enough, that's for sure.

After some updating I can play the files, but XviD as well as FFmpeg show
artefacts of motion at the top. That's bad, because then I don't know what
the problem is.
Anyway, I'll check, thanks for the clip.

gruel

P.S. Can you reproduce this with only 1  bframe? Without bframes? 
>From what I "see" I would guess, it shoulnd't matter, but who knows.
Can you reproduce without QPel? I would say, no, but please test.


On Sat, 9 Aug 2003, Radek Czyz wrote:

> Hi :)
> 
> Unfortunately, I have new examples of gmc not being decoded correctly
> on libavcodec. I just finished watching my newest encode and I saw
> this problem several times - it usually only applies to several
> macroblocks and happens in high motion.
> 
> This is the only good examples I found:
> http://homepages.ihug.com.au/~syskin/prob2.avi
> http://homepages.ihug.com.au/~syskin/prob3.avi
> 
> I can give you more, but it's usually very elusive, mostly because of
> high motion. It not always appears to be coming from the edges but
> maybe it does.
> 
> Unfortunately I can't compile ffdshow with some debug GMC output and
> compare against xvid's decoder :/
> 
> Regards,
> Radek
> 
> PS. I'm talking to Edouard, he says he's having problems both using
> xvid and libavcodec (first clip). I promise that my vfw devapi4 xvid
> build decodes both clips correctly. Bizzare.
> 
> _______________________________________________
> XviD-devel mailing list
> XviD-devel at xvid.org
> http://list.xvid.org/mailman/listinfo/xvid-devel
> 
>