[Ffmpeg-devel] VP3/Theora Perfection

Michael Niedermayer michaelni
Mon May 16 11:50:05 CEST 2005


Hi

On Monday 16 May 2005 09:17, Milan Cutka wrote:
> > I have been carefully reading the Theora spec and I think I understand
> > the loop filter now. Well, "understand" might be a strong word. At
> > least, I think I recognize the mathematical mechanics of the loop
> > filter. Unfortunately, my primary development machine is having troubles
> > so I can not effectively test the logic. I have attached a patch. Can
> > someone test it out and see if the video problems magically go away or
> > if the video at least looks a little better? Please test on a VP3 sample
> > since there is more logic to hook up for Theora.
>
> The patch didn't applied cleanly to CVS version of vp3.c:

patch -F9 will apply it automatically

>
> patching file `vp3.c'
> Hunk #8 FAILED at 2351.
> Hunk #9 FAILED at 2359.
> Hunk #10 FAILED at 2370.
> Hunk #11 FAILED at 2381.
> Hunk #12 succeeded at 2681 with fuzz 1.
> 4 out of 14 hunks FAILED -- saving rejects to vp3.c.rej
>
> After applying remaining changes manually and testing
> /MPlayer/samples/V-codecs/VP3/vp31.avi I saw no improvement.

ive found and fixed another bug in vp3.c (wrong dequant matrixes where used), 
but its still not artifact free, and this patch doesnt help much either

[...]
-- 
Michael





More information about the ffmpeg-devel mailing list