AV1 video codec finalised

[Camoxide]

https://www.guru3d.com/news-story/av1-video-codec-specification-presented-will-save-bandwidth.html

https://aomedia.org/the-alliance-for-open-media-kickstarts-video-innovation-era-with-av1-release/

 

AV1 is a royalty free video codec competing with the non-free HEVC/H.265 , despite being free it should offer a 30% improvement over HEVC. 

Quote

The final specification of the upcoming AV1 video codec has been finalized and published, it was announced by the Alliance for Open Media with partners like Google, Netflix, Facebook, Amazon and Microsoft. It can potentially replace codecs like HEVC and VP9, royalty free.

 

AV1 offers improved compression compared to vp9 or hevc, the video bandwidth reduction can run upwards to 30 to 40 percent, without you seeing a difference. The best thing yet, this is a royalty-free model. 

For nearly three years, the Alliance for Open Media (AOMedia) has been working in lock-step with its members, the world’s best-known leaders in video, to develop a better quality internet video technology that benefits all consumers. Today, the Alliance is proud to announce the public release of the AOMedia Video Codec 1.0 (AV1) specification, which delivers cross-platform, 4K UHD or higher online video, royalty-free – all while lowering data usage.

 

Should be great for all of you on data caps! 

Edited March 29, 2018 by Camoxide
Extra source

[Zodiark1593]

Crap, I don't want to swap all my gear again for a new codec, just after moving my stuff to HEVC. When Youtube went VP9, it killed the battery life on my older laptop 

[Trik'Stari]

Neat.

 

Maybe this means my ISP will stop throttling my youtube videos.

 

*doubt*

[dalekphalm]

I feel like this is relevant:

 

Unless you can guarantee full industry adoption, including DRM support (sorry folks, it's a reality of things like Netflix), then you're not gonna see good hardware support for this.

[Camoxide]

5 minutes ago, dalekphalm said:

I feel like this is relevant:

 

Unless you can guarantee full industry adoption, including DRM support (sorry folks, it's a reality of things like Netflix), then you're not gonna see good hardware support for this.

Three of the creators of this standard are Google, Netflix and Amazon so that should cover 99% of the streaming market. 

It has the backing of Intel, Nvidia and ARM so that covers the hardware side. 

[AshleyAshes]

But if there's no native hardware support, as in ASICs that can decode it in hardware instead of relying on a general purpose CPU, no one will use it.  I mean, geez, have you guys LOOKED at the CPU demands just for 4K@24hz CPU decoding?

[dalekphalm]

4 minutes ago, Camoxide said:

Three of the creators of this standard are Google, Netflix and Amazon so that should cover 99% of the streaming market. 

It has the backing of Intel, Nvidia and ARM so that covers the hardware side. 

This isn't entirely correct. The "creator" is the Alliance for Open Media. Google, Netflix, and Amazon are members, as are a number of other companies. That doesn't mean they're actually going to use AV1 in their commercial applications.

 

And even if they do, it's going to be a few years at best before you see even some hardware support.

 

When even HEVC has limited hardware support - only devices from the last couple years really.

 

Realistically, if they all decide to adopt the codec, we might see widespread support in 5+ years.

[DrMacintosh]

I think H.265 is good enough for now. We have too many video codecs. 

[BuckGup]

H.265 is fine I agree. Except no one excepts it properly except vimeo

[mariushm]

hevc didn't gain users due to licensing costs and multiple patent pools. only a few days ago they removed royalties for streaming (websites had to pay every time user watched videos) but there are still other problems making adoption slow. Guess they got a bit worried.

For same reasons you don't have hevc decoders in browsers (firefox has h264 decoder made by cisco which pays fixed fee as royalty, something like 5 million dollars a year , but i don't think hevc patent pools allow this or fees would be prohibitive)

 

av1 isn't that great.. imho quality per bitrate increases over hevc/vp9 are too small

[Taf the Ghost]

1 hour ago, AshleyAshes said:

But if there's no native hardware support, as in ASICs that can decode it in hardware instead of relying on a general purpose CPU, no one will use it.  I mean, geez, have you guys LOOKED at the CPU demands just for 4K@24hz CPU decoding?

It took a long time for the other Codecs to reach mass adoption. The more interesting question will be the compute load necessary to decode it. More compression normally means more decode resources necessary, but we'll see.

[themctipers]

4 hours ago, AshleyAshes said:

But if there's no native hardware support, as in ASICs that can decode it in hardware instead of relying on a general purpose CPU, no one will use it.  I mean, geez, have you guys LOOKED at the CPU demands just for 4K@24hz CPU decoding?

4k 60fps on youtube kills my CPU, but is completely fine on my laptop (because intel UHD can hardware accelerate it)

 

Let's hope this codec can be decoded hardware on existing chips or something, just needs drivers. 

[Qwertious]

With all the industry support that is backing AOMedia I don't see why this won't be widely supported within a few years. 

[Curufinwe_wins]

I would like to know how high the encoding/decoding cost for this new codec will be.

 

HEVC is something like 10x more demanding than H.264. It happens that cpu/asic capabilities have managed to keep up fine (and thus I am using x265 on everything I can atm), but I'm not very sold on another potential 2-10x computational cost leap for another 30% reduction in size.

[LAwLz]

10 hours ago, dalekphalm said:

I feel like this is relevant:

 

Unless you can guarantee full industry adoption, including DRM support (sorry folks, it's a reality of things like Netflix), then you're not gonna see good hardware support for this.

The members which has committed to supporting it are (emphasis on the ones people on here will know):

• Amazon
• Apple
• ARM
• Cisco
• Facebook
• Google
• IBM
• Intel
• Microsoft
• Mozilla/Xiph
• Netflix
• Nvidia
• Adobe
• AMD
• Amlogic
• Argon Design
• Ateme
• BBC
• Bitmovin
• Broadcom
• CableLabs
• Chips Media
• Hulu
• Ittiam
• NG Codec
• polycom
• Realtek
• Sigma
• Socionext
• Veri Silicon
• Vidyo
• VideoLAN
• Xilinx

Support won't be a problem.

 

Nvidia, AMD, ARM and Intel are already working on hardware support (the codec was designed in cooperation with them to make sure hardware support was efficient).

Full hardware support is on track to be finalized in 12-18 months, although it might take a bit longer before products start appearing with it. So by late 2019 or early 2020 we should start seeing it.

Until then both software and hybrid decoding will be available.

It is already supported in Firefox and VLC if you want to test it out.

 

 

 

10 hours ago, AshleyAshes said:

But if there's no native hardware support, as in ASICs that can decode it in hardware instead of relying on a general purpose CPU, no one will use it.  I mean, geez, have you guys LOOKED at the CPU demands just for 4K@24hz CPU decoding?

It will be used, and the CPU demands aren't that high. Don't get me wrong, they are not low, but they are not that high either. 

Both Google and Netflix has committed to be early adopters of it.

You can try it out here:

https://demo.bitmovin.com/public/firefox/av1/

 

Also, this is not just for 4K. One of the major benefits of AV1 is that it is much better than HEVC for lower resolution content.

 

 

9 hours ago, DrMacintosh said:

I think H.265 is good enough for now. We have too many video codecs. 

9 hours ago, BuckGup said:

H.265 is fine I agree. Except no one excepts it properly except vimeo

Quality wise HEVC is "fine", but who wouldn't want 30% smaller files?

Also, the big problem with HEVC is the licensing and patenting, which are absolutely atrocious.

 

 

5 hours ago, themctipers said:

Let's hope this codec can be decoded hardware on existing chips or something, just needs drivers. 

That won't be the case. New hardware is needed for full hardware accelerated decoding.

Hybrid decoding (for example shader based decoders) will most likely be supported on existing hardware, but that won't be nearly as efficient as fixed hardware.

 

 

4 hours ago, Curufinwe_wins said:

I would like to know how high the encoding/decoding cost for this new codec will be.

 

HEVC is something like 10x more demanding than H.264. It happens that cpu/asic capabilities have managed to keep up fine (and thus I am using x265 on everything I can atm), but I'm not very sold on another potential 2-10x computational cost leap for another 30% reduction in size.

The cost for encoders and decoders are 0 dollars.

It is completely free to use and the license for both the reference encoders/decoders and the codec itself even passes the Debian Desert Island Test (which means that anyone who downloads or uses it are automatically granted the rights to use them free of charge).

 

On top of that, as extra protection, all members have to submit their relevant patents to the alliance for free to use. If anyone sues anyone for using AV1, then that company or person suing automatically loses the rights to use all the other patents in the pool. So if Apple sues someone for using AV1, then all the patents from Microsoft, Google, Cisco, Nvidia, AMD, Intel etc could all be used to sue Apple back.

 

This is why AV1 is exciting. Not because it can save 30% of bandwidth (which is a nice bonus for consumers), but because it is finally a proper free and open video codec. Video codec licensing has always been an issue, and free codecs like Theora never got widespread adoption. This time that won't be an issue.

 

Edit: Just now realized you were talking about computational cost and not licensing costs.

AV1 is slow as fuck to encode right now, and pretty slow to decode too. But the format was not finalized until just now, so they have not spent any time improving speeds. That will be their focus from now on.

Ittiam will show a demo with a "significantly speeded-up version of AV1 encoder" on April 10. So performance will hopefully improve a lot very quickly.

Edit: I looked up the NAB schedule and apparently bitmovin's new encoder uses 6 times as little CPU power as the 2017 version. It should still be far behind for example x265, but that is a significant improvement nonetheless.

 

Some people in the alliance has also experimented with using AV1 for images (similar to what Apple does with HEVC in the HEIF container).

 

One of the major reasons why HEVC adoption was "slow", was because of a great deal of uncertainty about the patents (and there still is a great uncertainty).

Right now, there are three different HEVC patent pools which you may or may not have to pay royalties to. On top of that, there are several individual patents in it, whose owners have not yet announced how much they want in fees. It's a clusterfuck so colossal that even Microsoft and Apple decided that enough was enough, despite them pushing so hard for AVC to be the HTML5 standard just a few years ago.

 

 

Edit 2:

False alarm everybody. The bitstream is not yet frozen, but it soon will be. Seems like the marketing guys av AOMedia were a bit too trigger happy. It should be frozen very soon though.

Quote

It's not even a bitstream freeze. This 'release' was put out by the marking folks, and wasn't even discussed with people on the AOM list (I'm part of AOM via VideoLAN). The bitstream remains under development.

Near as I can tell this is just a PR piece before NAB.

 

[Mihle]

Smaller file size and the same image quality?

I don't care.

 

Bigger file size and better image quality?

I do care. 

 

Same file size but better image quality?

I do care.

[BuckGup]

3 hours ago, Mihle said:

Smaller file size and the same image quality?

I don't care.

 

Bigger file size and better image quality?

I do care. 

 

Same file size but better image quality?

I do care.

You sorta contradicted yourself. You could have the same file size with better image quality because it's smaller then before

[bcredeur97]

12 hours ago, BuckGup said:

H.265 is fine I agree. Except no one excepts it properly except vimeo

I use it for media streaming with my Plex server at home. it works great.

But 4K videos destroy my 3570k... though I literally have a 3960X in the mail and should arrive today.

[mariushm]

RX 5xx series and GT 1030 / GTX 10xx have hardware HEVC decoder (at least up to 4K 60 fps) . They also have hardware HEVC encoder (though the RX 5xx series will do only 4K 30fps in real time or 1080p 60fps while the GT 1030 has hardware encoders disabled, so you need at least GT 1050.. at least I think HEVC encoder is still enabled on GT 1050)

 

So maybe it's time to upgrade your video card instead of the CPU?

 

[Misanthrope]

36 minutes ago, mariushm said:

RX 5xx series and GT 1030 / GTX 10xx have hardware HEVC decoder (at least up to 4K 60 fps) . They also have hardware HEVC encoder (though the RX 5xx series will do only 4K 30fps in real time or 1080p 60fps while the GT 1030 has hardware encoders disabled, so you need at least GT 1050.. at least I think HEVC encoder is still enabled on GT 1050)

 

So maybe it's time to upgrade your video card instead of the CPU?

 

What about the Ryzen 2200/2400g? I imagine they do but they'd be ideal for media centers.

[mariushm]

Those are slightly tweaked Vega graphics ... pretty much same encoding/decoding engine as the RX series , so yeah ...should be this or better:

4k h264 up to 120fps or more, 4k 60 fps decoding with HEVC and semi-hardware decoding of VP9 (hardware accelerated but some decoding done on cpu/software).

4k 60fps on h264 encoding, 4k 30fps guaranteed hevc encoding, can do more fps depending on quality settings but 60fps I'm not sure it's reliable.

[Mihle]

3 hours ago, BuckGup said:

You sorta contradicted yourself. You could have the same file size with better image quality because it's smaller then before

Not really. My post was more ment at how a codec is used by companies rather than how the codec itself works. So what you say there is exactly what I ment that I want the companies to use the codec for, but probably not what they are going to do.

 

Companies would most likely use the codec to have the same noticeable quality for less size.

 

While I want them to rather use it to get better noticeable quality out of the same size.

 

I also wish the bitrate of like YouTube and Netflix was just higher to get better noticeable quality.

[LAwLz]

25 minutes ago, Mihle said:

Not really. My post was more ment at how a codec is used by companies rather than how the codec itself works.

 

Companies would most likely use the codec to have the same noticeable quality for less size.

 

While I want them to rather use it to get better noticeable quality out of the same size.

 

I also wish the bitrate of like YouTube and Netflix was just higher to get better noticeable quality.

Youtube's problem is not really the bit rate. Youtube actually has a pretty high bit rate.

The problem is the settings they use. In order to get decent speeds they are using looser settings, which they have to because 300 hours of video needs to be encoded every minute, and each video is encoded to several formats.

 

You can get far better looking videos at lower bit rates than Youtube uses, even with AVC. The problem is that it takes maybe 20 times as long to encode.

[yian88]

Meh just the spec, this is just the beginning, did they implement fast high quality encoders decoders multithreaded?no, others have to do it and it takes forever. There is also no GPU decoding for it 8/10/12bit 4k60 hdr/10, without that this codec is meaningless for many years to come.

30% More efficient i assume 30% less size on disk and bandwidth required for stream, but that means nothing if the decoding performance is non-existent like hevc 4k 10bit ~50mbps, your PC will burn down with choppy video if the GPU doesnt support  it(my 970gtx doesnt and the cpu cant handle it).

[MyName13]

Did VP8 and VP9 get any support?