AOMedia Announces Public Release Of AV1 Video Format

[torsionbar28]

Sorry to disturb your dream and lower your expectations, but have you read the list of invovled enterprises? It reads like a collection of who is the nastiest? (yeah, spare me with Mozilla...)
...
Honestly, I would not expect much battle here. And maybe Mozilla will deliver a non-DRM variant officially, that is only capable of playing "elephant's dream" as the sole video in this codec world-wide, while all others will have the DRM variant and content will be nailed down with DRM.

Sorry for not being more optimistic, but I have >25 years of experience with that stuff and it would be a (pleasant) surprise if it was different this time.

^ With you 100%, with all those names signing on. It's a guarantee that this "open" standard will come with a mega DRM infection.

Also, where does this leave VP9? I thought VP9 was supposed to be the answer to a free and open H.265 alternative?

[Gusar]

For those going on about DRM... You should know that DRM is not applied at the codec level, but at the container level. Talk about DRM when it comes to AV1 is irrelevant. Sure, a lot of companies that'll adopt AV1 will put it in a container wrapped in DRM, but that still doesn't affect AV1 itself in any way. AV1 is free to use as you please.

About the question of where this leaves VP9, that's obvious, AV1 is the successor to VP9. In fact, the AV1 reference implementation is a continuation of the libvpx (Google's VP8/VP9 implementation) codebase.

[karasu]

For those interested I made a small comparison with AV1 VP9 x265 and x264 at very low bitrate. Spoiler: AV1 is sloooooooooooooow

http://video.1ko.ch/codec-comparison/

[DrYak]

Also, where does this leave VP9? I thought VP9 was supposed to be the answer to a free and open H.265 alternative?

VP9 *was indeed* supposed to be the answer to h265.

VP10 was supposed to be the answer to whatever comes next.
Combined with Xiph's Daala and Cisco's Thor which each started with the same target (being one generation further/better than h265), that what eventually evolved into AV1.

Basically VP9 is to h265/hvec what AV1 is to JVET, whenever that things ends up being announced by MPEG (supposedly expected in 2020, with 30-50% performance improvement - basically the same range that AV1 reaches today. meh.)

But given the massive influx of brain power behind AV1, AOMedia managed to beat them today (again, at the same efficiency that is planned for JVET).

[cl333r]

How exactly does AV1 achieve 25+% better efficiency over VP9 & HEVC?

[bemerk]

For those going on about DRM... You should know that DRM is not applied at the codec level, but at the container level. Talk about DRM when it comes to AV1 is irrelevant. Sure, a lot of companies that'll adopt AV1 will put it in a container wrapped in DRM, but that still doesn't affect AV1 itself in any way. AV1 is free to use as you please.

It is relevant because if you use Kodi on your HTPC and the DRM can only be used with software decoding due to architecture limits than it becomes important how fast your cpus have to be to decode it in realtime for 720p,1080p or even 4k material. Hardware decoding won't help at all there as it can't be used.

[Brisse]

How exactly does AV1 achieve 25+% better efficiency over VP9 & HEVC?

As always when a new more efficient codec comes along, it throws more complex math at the problem, which means it's going to be slooooooooow. 50% increase in compression ratio for a lossy video codec usually comes at a price of 10-20 times longer encoding times.

[starshipeleven]

Sorry to disturb your dream and lower your expectations, but have you read the list of invovled enterprises? It reads like a collection of who is the nastiest? (yeah, spare me with Mozilla...)

Does this make AV1 any less open and any less free?

Can I remind you that it was Google + Microsoft + Apple + Mozilla that more or less ended the browser wars by sitting at a table and coordinating a new standard (HTML5)? Does the companies involved make that accomplishment any less good?

IBM = lolwut, I thought they were dead
(once, back in the days of mechanical keyboards, this was a tech invention force, but now, what are they now? The probably don't even know for themselves!)

IBM is still fine and still good on opensource, maybe you lived under a rock?
For example they are the one and only high-performance CPU manufacturer that provides fully open firmware and microcodes for their boards, look up OpenPower project, or Talos II workstation for a relatively affordable board with Power CPUs.

Honestly, I would not expect much battle here.

The battle here is just nuking HEVC and replacing it with AV1 and yes I see them putting up a fair fight if they all start pushing AV1 down people throats on their own channels (youtube, facebook, netflix for example).

The big step is ensuring that all hardware has AV1 hardware decoding, if that happens then the ball will start rolling for everyone else too, as it is a no-brainer, no license fees, no legal bullshit.

I have >25 years of experience with that stuff

I'm pretty sure there isn't much precedent to this. I don't remember other open formats like Vorbis having more or less unanimous support from every major PC and internet-based company, where many have announced that they will push it down people's throats.

[starshipeleven]

^ With you 100%, with all those names signing on. It's a guarantee that this "open" standard will come with a mega DRM infection.

100% ignorant bullshit like his. AV1 is a codec, an algorithm/standard to compress raw video into a file, and being open and free means that everyone can compress his media streams with it without facing patent trolling and paying license fees (and risk big costs down the line when the patent holder starts rising the prices for the license after you can't back out anymore).

DRM is related to the freedom of the content of the media file, which is not related in any way to the compression algorithm.

[starshipeleven]

How exactly does AV1 achieve 25+% better efficiency over VP9 & HEVC?

It's an order of magnitude slower when encoding, for starters.