Quality comparison

[colleenkhenry]

I see there are no claims as to quality per bit here. I understand that quality is a super hard thing to measure and it depends on a ton of factors, however, I'm curious if you could throw out some simple claims.

For example, you could pick some clips from the derf set and say "according to VMAF we are X better than x264 medium mode with Y GOP size and Y bit rates and Z resolutions."

I'm thrilled for the contribution to the community, don't get me wrong, but, that sort of data might help people get over initial skepticism that pervades this industry (and rightly so) when it comes to encoding.

[1480c1]

One thing that is possible is to use a well known clip such as big buck bunny and jelly fish and iterate through the tests and do vmaf, psnr, or ssim with x264, x265, and libaom along with bit rate comparisons. It would still be hard to put together the data unless there is someone good at that kind of thing.

[colleenkhenry]

@chadnickbok I think had some interest in doing the tests if someone tosses him the data set

[tianjunwork]

@colleenkhenry Thank you for your suggestion. Yes, as you said, some benchmark document is needed to publish. The team is all busy on AV1 now. But we welcome any contribution from the community.

[colleenkhenry]

Go go go on AV1! Don't let me slow you down ;).