[FFmpeg-devel] donation for snow
Fri Nov 7 04:07:14 CET 2008
On Thu, Nov 6, 2008 at 4:39 PM, Baptiste Coudurier
<baptiste.coudurier at gmail.com> wrote:
> Jason Garrett-Glaser wrote:
>>> Do you have any comparision data ? I would, too, be interested in having
>>> these comparisions, also knowing which jp2k encoder was used, test
>>> material, etc...
>> I remember being linked to a paper on the topic a while back which had
>> the numbers I was using (though I don't remember them exactly).
>> However, I don't trust PSNR anyways, and if you wanted to do a real
>> comparison, I would say that a visual test would be the only one worth
>> bothering with.
> I understand, what puzzles me is that D-Cinema guys, who are know to be
> really picky about picture quality (choosing 12 bit XYZ is a thing),
> chose jp2k over other codecs, having done many extensive visual tests
> (they don't trust PSNR, Im not sure they care about it).
That's odd. My general experience with wavelets is that they give
much worse "visual quality per PSNR" than DCT-based does. That is,
you can get many cases where a fully PSNR-optimal wavelet image is 3db
PSNR better than a fully PSNR-optimal DCT image (e.g. JPEG2K vs JPEG)
but the DCT-based one *still* looks better.
More information about the ffmpeg-devel