The Alliance for Open Media - AV1 Format. Any thoughts.

[LoLDrifter]

I know I am jumping the gun here, but https://bitmovin.com/bitmovin-av1-encod ... streaming/ They have said to show off an "AV1 Livestream that will show 1080p live broadcast quality at just 1.5 Mbps" at NAB. Also this line "Currently you would need around 8 to 15 Mbps with traditional codecs like H264 to deliver the same quality. With AV1 you can reduce your CDN and storage cost by up to 10x."

I would love to be an early adopter of this video format, especially since it seems to be coming built in to html5 unlike hevc. It seems to be the next big jump in video encoding but we still hear very little about it. I would love to know if the HandBrake team has been experimenting with encodes, if so how is it going. Can you imagine the amount of space we can save in the future.

Anyway I am super excited about this new format would love to know more. Especially when you guys think we might see this, hoping as soon as next year.

[Deleted User 13735]

Thanks for the link. A year? You'd better get started coding right now.

[s55]

You really need to dive into this more than looking at high level marketing.

So, for instance, notice they compare against H.264 and not H.265 or even VP9. Why? Educated guess but there are probably 2 main reasons.
1. They are marketing to people using H.264 today
2. The picture is no-where near as rosy when compared to H.265 or VP9

The second claim they make "Currently you would need around 8 to 15 Mbps with traditional codecs like H264 ".
OK, How was it encoded? What encoder was used? What type of content was used?

I'd also make the point that all encoders are made equal. http://www.compression.ru/video/codec_c ... esults.png AS you can see there is a massive difference in encoders.

I'm currently watching a variable bitrate H.264 DVB stream that's averaging about 4.5~5mbit. Even with high motion it's only peaking at around 8~9 mbit for short periods.

The point I'm making here, is you really need to delve deep into these codecs and encoders and understand your own use case scenario and make the best choice so you can optimise for your use case. Just quoting the the high level marking will mis-lead you.

Are you currently using H.265 or VP9? If not, why not? They both offer large advantages over H.264 in a lot of common use cases.

As we did with VP9, we'll evaluate AV1 when the encoders start to mature. I'm defiantly not expecting that to happen for a good 2+ years. The x265 encoder is quite usable but VP9 is still horrifically slow so it's going to be an uphill battle for AV1 here. The AV1 bitstreams not due to be finalised until the end of this year last I checked and the folks behind the encoders need substantial time to optimise.

[LoLDrifter]

Thanks for the insight s55, I will slow my roll. Was probably reading into this with all these backers "Amazon, Cisco, Google, Intel, Microsoft, Mozilla, Netflix, AMD, ARM, NVIDIA and bitmovin." made me feel like well [Censored] is going down with this format. If these companies as a group can't figure out how to make the best video format no one can.

I do love Hevc, but it seems like it will never reach the point that H.264 has when it comes to online streaming or generally adopted across all devices. Especially since this Alliance does not want to pay them so much that they are making their own.

[mduell]

Thanks for the insight s55, I will slow my roll. Was probably reading into this with all these backers "Amazon, Cisco, Google, Intel, Microsoft, Mozilla, Netflix, AMD, ARM, NVIDIA and bitmovin." made me feel like well [Censored] is going down with this format. If these companies as a group can't figure out how to make the best video format no one can.

I disagree -- any one of them could do better than they will do collectively. Death by committee. They'll never deploy anywhere near the same time, if at all. It's just politics against MPEG LA, same hype as VP8 or VP9 or WebM or whatever.

Also a 10x improvement is 100% marketing [Censored], they'll never achieve that in any reasonable test vs H.264 with a decent encoder much less H.265.