[MPlayer-users] Re: [Ffmpeg-devel] lavc mpeg-4 vs. VP6 using On2's raw source files

[Corey Hickey]

D Richard Felker III wrote:
> On Wed, Nov 05, 2003 at 03:37:34PM -0800, Corey Hickey wrote:
> 
>>I was curious about how well lavc compares to VP6. All I could really
>>test was PSNR, but the results are favorable.
>>
>>http://fatooh.org/vp6-lavc/
>>
>>Please tell me if I've screwed anything up. :)
> 
> 
> Quite amusing, but not unexpected. :)
> 
> BTW, if the VP6 white paper isn't lying, the H264 implementation they
> used must totally suck. A proper H264 should do better than lavc's
> mpeg4, not worse than VP6. Also I noticed in the pictures they show
> for comparison with WMV9 and H264, the artifacts in both of the latter
> could easily be corrected with elementary postprocessing. I suspect
> VP6 does its own sort of postprocessing internally so they can try to
> claim better quality. IMO a fair comparison of video codecs is not how
> smooth versus blocky they end up looking, but how much detail is
> retained or lost.
> 
> BTW, could you add divx5 and xvid to the tests? :)))))))
> 

Certainly, except...

When attempting a two-pass encode with divx5, I get the problem
described in this thread:
http://www.mplayerhq.hu/pipermail/mplayer-users/2003-July/034683.html

The answer then was, of course "don't use it", but that also means I
can't do any two-pass encodes. The URL above doesn't have a verbose log,
so I attached one. If you want a proper, detailed bugreport tell me and
I'll write one when I have time (tomorrow).

Also, do you know how to get the PSNR out of a divx5 encode? I don't see
a "psnr" option, and don't know any other way.

Xvid looks like it'll be no trouble.

-Corey
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: divx.log
URL: <http://lists.mplayerhq.hu/pipermail/mplayer-users/attachments/20031105/85e4c71d/attachment.asc>