[XviD-devel] Result of ffmpeg encoding tests

[Michael Militzer]

Hi,

----- Original Message -----
From: "Christoph Lampert" <chl@math.uni-bonn.de>
To: <xvid-devel@xvid.org>
Sent: Thursday, January 16, 2003 10:43 AM
Subject: [XviD-devel] Result of ffmpeg encoding tests


> Hi,
>
> our collegues of ffmpeg got a great overview of tests about PSNR and fps
> for various combinations of their option. We should do something similar
> for XVID... and also, we should "steal" the features from ffmpeg which are
> most beneficial, e.g. "vhq" mode decision (based on real number of
> bits needed for a macroblock, right?) But maybe ChromaME has a similar
> effect?
>
> http://rguyom.chez.tiscali.fr/libavcodec_tests.html
> http://rguyom.chez.tiscali.fr/part1.html
> http://rguyom.chez.tiscali.fr/part2.html
> http://rguyom.chez.tiscali.fr/part3.html
> http://rguyom.chez.tiscali.fr/part4.html
> http://rguyom.chez.tiscali.fr/part5.html

yes, we should have something similar. It would be cool if we could agree on
a common suite of test sequences and maybe a benchmarking script which would
allow automatic testing of new features.

Also I quickly tested ffmpeg's MPEG-4 encoding capabilities earlier this
week and had similar impressions: 4MV mode seems to give no improvement in
ffmpeg if used without vhq. XVID also does not benefit much from 4MV mode
currently, I wonder if this could be improved by implementing something
similar to vhq mode decision.

bye,
Michael