next generation video: Introducing AV1, part1: Chroma from Luma

[xiphmont]

My first technical writing regarding the new AV1 codec is up at Xiph.Org. We've been working on AV1, heads-down, for a long time and my writing took a hiatus for almost that entire period. Now that it's out, it's time to continue the technology pages that I started with Daala.

"AV1 is a new general-purpose video codec developed by the Alliance for Open Media. The alliance began development of this new codec using Google's VPX codecs, Cisco's Thor codec, and Mozilla's/Xiph.Org's Daala codec as starting point. AV1 leapfrogs the performance of VP9 and HEVC, making it a next-next-generation codec. The AV1 format is and will always be royalty-free with a permissive FOSS license."

Comments

Promising codec

[imasin]

I haven't truly been interested in a lossy video codec since I came across the SIF1 codec back in 2011. Most of my personal video archive has been encoded in SIF1 because I have found that it handles grainy and noisy sources far better than any other codec I know of. H.264 simply chokes on high grain sources and the artifacts from HEVC sometimes look even worse than H.264.

SIF1 is barely past initial development and yet it still compares favorably with x264, the best H.264 encoder available. If AV1 can outperform SIF1 I may have found a new archiving codec, at least until the day storage becomes so cheap I can store video losslessly like I do with audio.

[mzso]

Since they take up so much space were there no attempts on dumping I-frames?
It seems to me that any compression improvement that can be gained from using information from following frames would help somewhat at least. (Completely abandoning it probably a lot more)

PS:
It's only possible to comment with a dreamwidth account now because the captcha is broken.

[xiphmont]

Without I-frames, you have no starting point, ie, there are no subsequent frames. The reasoning becomes circular.

For the record, analysis of later frames can (and do) inform the encoding of an I frame, but the content of the I-frame itself doesn't depend on any other frame.

Sorry to hear about the capcha, hopefully Dreamwidth fixes it. Without captcha (on livejournal) I was deleting hundreds of spam a day.

[mzso]

How so?
I remember you mentioning experimenting with GOP-s that didn't have an I-frame.

I guess information to be referenced needs to be encoded somewhere, but is having the first frame fully encoded without using any information from others ideal for compression?

What happens when there's a scene cut in the middle of a GOP?