An Open Source, Royalty-Free AV1 Codec Has Been Released (aomedia.org) 104
Artem Tashkinov writes: After three years in development the Alliance for Open Media is releasing the royalty-free AOMedia Video Codec 1.0 (AV1) specification. The AV1 codec promises an average of 30 percent greater compression over competing codecs according to independent member tests.
The release of AV1 includes:
The release of AV1 includes:
- Bitstream specification to enable the next-generation of silicon
- Unoptimized, experimental software decoder and encoder to create and consume the bitstream
- Reference streams for product validation
- Binding specifications to allow content creation and streaming tools for user-generated and commercial video
So, how long before... (Score:1)
Re:So, how long before... (Score:5, Informative)
There are all SORTS of patents on this codec. But everyone patent owner has agreed (and signed) to make those patents available royalty-free to the AV1 group.
Re:So, how long before... (Score:5, Insightful)
Re: (Score:2)
Adding x to y patent doesn't make it novel and thus would invalidate the combo x and y patent. Derivative works are not patentable.
Re: (Score:1)
Re: (Score:3)
Re:So, how long before... (Score:5, Informative)
It might happen, but it hasn't happened with VP9. It also hasn't happened with Opus.
Furthermore the AV1 license is structured so that if you sue someone for using AV1, you lose your own rights to use AV1. Thus, only pure-troll entities will be able to initiate such lawsuits. That limitation didn't apply to previous codecs.
Re:So, how long before... (Score:4, Interesting)
Furthermore the AV1 license is structured so that if you sue someone for using AV1, you lose your own rights to use AV1. Thus, only pure-troll entities will be able to initiate such lawsuits
That's not a real problem. The normal way for such lawsuits is to spin out a company that does nothing other than own the patents and license them back to the original company. That company can then sue everyone, but doesn't do anything other than license patents so is immune to countersuits.
Re: (Score:2, Informative)
They can try. But with 30 companies doing IP Review, I doubt they will.
Re: (Score:2)
We get patent lawsuits on some of the worlds largest countries who take Patents very seriously. Find some obscure patent troll had made some lame patent in the past that no one would think of even bother looking for because it is so obvious that only a moron would try to Patent it and a moron patent official would approve it.
Re: (Score:3)
You've got the 1st (Apple), 2nd (Alphabet/Google), 3rd (Microsoft), 4th (Amazon), and 6th (Facebook) most valuable company in the world on the list, plus countless others in the top-50 list. Their lawyers are plentiful to support AV1. (note this list was quickly pulled from a source that is about a year old, I know their rankings have slightly shifted since then, but the main fact still remains that they're the powerhouses of the entire world now)
Re: (Score:1)
C'mon man, almost all of us here can count to 50.
Re: So, how long before... (Score:1)
Apple was not a founding member by they joined as a governing member some time ago.
patent troll comes calling (Score:2)
... like say... RAMBUS?
Re: (Score:1)
Spreading fear, uncertainty, and doubt about AV1 is a tactic HEVC patent pools have started to use. Velos Media says it's a nice codec you've got there, it'd be a shame if something happened to it [cnet.com].
Re: (Score:3)
some obscure company (patent troll) claims this violates all their patents?
if you're that paranoid then best stick with MPEG-2 and MP3 until the patents on AV1 and Opus expire.
Re: (Score:2)
Re: (Score:1)
How is it useless? You can easily decode it via software.
Re: (Score:2)
Which requires a powerful and power hungry CPU, probably complete with active cooling...
Mobile devices won't be able to do it, or won't do it for long before the battery dies.
TVs and STBs won't be able to do it as they typically contain low power general purpose processors, one containing a powerful general purpose processor would require active cooling and be noisy - not what you want while trying to enjoy a movie.
It will take a while before this will see much adoption, although its a welcome start.
h265 (Score:5, Informative)
Which requires a powerful and power hungry CPU, probably complete with active cooling...
Mobile devices won't be able to do it, or won't do it for long before the battery dies.
Which is *also* the case with h265, mostly due to the patent minefield and high licensing costs causing lots of hardware manufacturer to backtrack on their intentions to feature dedicated h265 cores in their latest hardware.
Which is the whole reason Daala, VP10 and the other pieces of what eventually combined into AV1 were started in the first place.
Except for a few select phone (Apple's iPhone) lots of (cheaper) mobile devices haven't started getting real h265 decoding, neither.
And again, AV-1 was designed to be GPGPU friendly.
Re: (Score:3)
If I'm watching movies on my i7 I don't really care if I'm using a large fraction of my CPU performance in software decode since I'm not really doing anything else with my computer. (I don't normally watch a movie while playing a game and typing up my thesis)
For STB, you could probably still do software decode on a more powerful one. I haven't tried it yet, but Nvidia Shield might have enough grunt to do the job at 1080p. (I doubt 4K, but I don't have a 4K TV yet)
Re: (Score:1)
There's more to life than PCs. Try decoding h.265 on a phone or tablet that doesn't support hardware decoding. On my three year old Sony tablet, playing back high definition h.265 movies in MX Player with software decoding, the frame rate is unwatchably low, and I hate to think what it does to the battery life. I have to make sure when selecting files for a trip, that I stick to h.264. There's no reason to think AV1 will be any easier to decode in software than h.265. Given that mobile is, you know, quite i
sort of hardware decoding (Score:5, Interesting)
This is awesome, but it's useless until decoder hardware is prevalent.
The plan is that the initial quick deployment of it will rely on shader code, so decoding will be hardware accelerated, but GPGPU instead of dedicated hardware code.
On the other hand, you have a bunch of hardware manufacturer on the board too (dedicated hardware manufacturer like Broadcom, GPU manufacturer AMD, ARM, Intel, Nvidia) and they have been taking part in the process, guiding selection of some technology (the reason why ANS was dropped in favor of Daala_ec, as it's more hardware friendly, etc.)
They have probably already started testing hardware implementation while the development process was going on. So maybe AV1 spetialized decoding cores might show up faster than expected.
Unoptimized (Score:1)
Judging from other open source software, that will be a show stopper. Having millions of eyes on the code does not automatically cause the owners of those eyes to give up their day job to hunker down and turn the code into something productive and efficient.
Re:Unoptimized (Score:4, Insightful)
What are you talking about? The opensource implementations of MP3 and H264 quickly outpaced every commercial implementation, I would expect the same for this codec. What they won't be able to do is compete with embedded hardware implementations that will come but history has shown that the opensource version will be less efficient but more feature rich than those hardware implementations.
Re: (Score:2)
LAME was indisputably better than Fraunhofer by 2001. The project started in 1998, 3 years to beat the company that did most of the research behind the standard is pretty damn quickly.
Re: (Score:2)
LAME did this by infringing on patents and copyrighted code.
It's called "Lame Ain't an MP3 Encoder" because of this. They hid behind the claim that they're only doing this for research and are only creating source code, not an encoder. Of course, we all know LAME is in fact an MP3 encoder. All you have to do is compile it or download a compiled binary from someone else who did that step for you.
See Microsoft, DivX, and XviD for a similar situation in video.
Re: (Score:2)
LAME was indisputably better than Fraunhofer by 2001.
LAME did this by infringing on patents and copyrighted code.
Grandparent said 2001. LAME had already removed all of ISO's copyrighted demo code by May 2000. The patents were an issue only because compression users hadn't yet begun to push toward royalty-free codecs. By that time, Xiph's Vorbis encoder was handily beating both LAME and Fraunhofer's MP3 encoder in rate-distortion performance.
Re: (Score:2)
Yes, but the argument was that the opensource development model can't make a superior implementation quickly, that is demonstrably false and patents and copyright will not be an issue with this open royalty-free standard.
Re: (Score:1)
It's not demonstrably false. They haven't so far shown that they can do that. The major successes so far have all infringed on patents and stolen code, even if they later clean up the act to remove infringing bits.
LAME, DivX ;-), XviD, x264, and x265 are the major successes of open codecs so far. (Though DivX ;-) became DivX, was bought by Fraunhofer, wasn't fully open/free, etc., its origins are still in the same camp.) All of these have their roots in the tainted swamp of code theft and patent infring
Re: (Score:3)
What are you talking about? The opensource implementations of MP3 and H264 quickly outpaced every commercial implementation,
Really? I used the commercial Fraunhofer MP3 implementation for almost a decade, because the open source Lame was both more resource hungry and gave more artefacts. That's not my definition of "quickly".
Re: (Score:2)
Also, LAME was always better at the bitrates that quality audiophiles care about.
No, it wasn't. In the early days, lame was all about VBR, which no audiophile would touch with a conductor's pole. VBR sounds shitty, because of the quality changes being so readily detectible. A change back and forth between good quality and very good quality is more perceptible than a track that stays at mediocre quality.
For 320 kbps, Fraunhofer had the upper hand, especially for encoding quality (you can tell whether an old MP3 uses FH or lame based on whether high hats sound like being stricken by a
Some caveats (Score:5, Informative)
As with VP9 earlier, the first reference AV1 encoder is absolutely slow: currently it's an order of magnitude slower [doom9.org] than x265's veryslow preset (which is extremely slow to begin with).
AV1 is not currently supported by anything under the sun except an alpha build of Firefox [mozilla.org] (where it struggles to decode even a 3Mbps video on powerful PCs).
Most likely ffmpeg will include its own decoder (implementation) because ffmpeg and AV1 developers have contradicting views on coding styles. ffmpeg has its own VP9 decoder.
Apple joined [macrumors.com] the alliance just a few months ago when the development was almost over, which means Apple most likely didn't really contribute to it at all.
The spec is 619 freaking pages long [github.io].
Re:Some caveats (Score:4, Interesting)
As with VP9 earlier, the first reference AV1 encoder is absolutely slow: currently it's an order of magnitude slower [doom9.org] than x265's veryslow preset (which is extremely slow to begin with).
At 40% better efficiency than x265, slowness is a given and perfectly forgivable. For all intents and purposes it's like a next generation codec, but license free. Now it "just" needs the hardware to enable it as a viable choice.
Apple joined [macrumors.com] the alliance just a few months ago when the development was almost over, which means Apple most likely didn't really contribute to it at all.
I don't see it being a problem except maybe for Apple. The fact they joined shows they reckon its value and I guess it's more than enough "contribution" at this point.
Re: (Score:3)
VP9 is partially hardware accelerated on Broadwell+, fully on Kaby Lake+, and fully with either major GPU platform with recent GPUs. If you've got legacy equipment and are trying to multitask while Youtube videos play then I guess the difference might matter, otherwise why would you care?
Re: (Score:1)
Haswell or Radeon 7970 or xyz_hardware aren't exactly legacy equipment. These days "legacy" hardware can be an order of magnitude faster than brand new hardware, or the other way around, depending on how high end or cutting edge was what you bought five years ago, or depending on the hardware's power budget.
In a way things are like they were in the 80s or early 90s : an Amiga 500 wasn't legacy in 1992, an IBM PC or Apple II or Commodore 64 wasn't legacy in 1987.
Re: (Score:3)
VP9 is hardware accelerated in Firefox but only if you're on Windows 10 and have an appropriate CPU/GPU: Intel Skylake or better; NVIDIA Maxwell Gen2/Pascal; AMD Polaris/Vega ;-)
I'm not aware of other web browsers but Chrome and Edge might support hardware decoding on Windows 10 as well.
As for Linux it's all very sad - even H.264 is not hardware accelerated.
Re: (Score:2)
As for Linux it's all very sad - even H.264 is not hardware accelerated.
Hardware decoding of H.264 is supported under Linux through VA-API on Intel GMA 4500, Ironlake Graphics and newer, and AMD Radeon HD 4000 and newer. It is also supported through VDPAU on AMD Radeon HD 4000 and newer and nVidia GeForce 8 and newer. There are adapter libraries available in case you need to use a VA-API client with VDPAU drivers or vice-versa. Accelerated encoding is also supported on certain hardware.
The above is based on the Hardware video acceleration [archlinux.org] page on the Arch Linux wiki, and my own
Re: (Score:2)
I was talking specifically about web browsers (since the original poster was talking about the difficulties of playing VP9 content on youtube). I'm perfectly aware that Mplayer/ffmpeg has had HW decoding acceleration for many years already - in fact I use VDPAU all the time.
Re: (Score:2)
Does VP9 hardware acceleration automatically include WebM? I'm assuming so because their web site still mentions VP9 all over the place years after WebM came out.
Example:
https://www.webmproject.org/ha... [webmproject.org]
Re: (Score:2)
Ummm, nevermind. I'm conflating containers and codecs.
Re:Some caveats (Score:4, Informative)
Damn, give it some time. It was just released. When looking at things like this, you have to look at the backing and support it gets.
Give it a year or so, and see where we're at. This codec is here to stay, and while the encoding speeds are slow now, it has an enormous backing to win in the end.
Preliminary support (Score:4, Informative)
VLC has also preliminary AV1 support (since 3.0).
G-Streamer has too (since 1.14)
Bitmovin has started offering experimental AV1 cloud compression for quite some time.
And given the long list of companies involved, more is going to come any moment soon.
Re:Some caveats (Score:4, Informative)
Who cares when Apple joined, as long as they did. That's a big pool of patents that they've owned for a long time that became available to the group, and some very deep pockets that can help defend (or at least won't sue).
With Apple, Microsoft, Google, Intel, Nvidia, AMD on board, it's a done deal. In 18 months it will be difficult to find a shipping device that doesn't support AV1 at some level.
Fuck MPEG-LA. Couldn't happen to a nicer bunch of assholes.
Re: (Score:2)
Sure but that reference encoder isn't even optimized. People are going a little crazy assuming this will need dedicated hardware to become useful when it was designed to be easy to optimize and has not yet been optimized. Bitching because the brand new shiny and cool codec isn't as mature as the codecs of 5-6yr
GPGPU (Score:2)
The plan of AOMedia was to shift the work to the GPU as soon as possible (think Vulkan/OpenCL compute shader)
And given the names on the list (several hardware manufacturers), you can bet that dedicated hardware AV1 cores are going to come next after that, much sooner speed than you would otherwise expect.
Re: (Score:3)
crawl -> walk -> run
I'm pretty sure Nvidia, Intel, and AMD have something to say about that.
AOMedia has the math working, and a published spec. Now they go over the code and make it faster, and work with the hardware guys. With the general hate for MPEG-LA out there, I wouldn't be surprised if in three years we look back on H.264 the way we look at MPEG-2 today. Man, that was a great standard back in the day, but the files are HUGE compared to AC-1!
Web (Score:2)
it's targetting web.
on the web, DRM is handled at another level (EME - encrypted media extensions).
video codecs are orthogonal to it.
basically, content providers don't pay attention if AV1 supports DRM or not (it doesn't).
What interests them is if the browser supports widevinecdm (it does if you browser can play netflix).
MPEG (Score:1)
How badly did they screw up for things like this to happen. Making the competitors (Amazon, Google, Apple, Netflix, etc) come together to make a replacement.
Re:MPEG (Score:4, Informative)
So badly that the chairman of MPEG says HEVC is unusable [chiariglione.org].
Re: (Score:2)
"Making the competitors (Amazon, Google, Apple, Netflix, etc) come together to make a replacement."
Gives a very strong clue as to why free licensed solutions continue to get development. They just get developed by funding from people who actually need the solution rather than a middleman who wants to build it and license to all those people.
Standards (Score:4, Funny)
Re: (Score:2)
"14 competing standards"?
I guess you could say AV1 came into being because HEVC was failing to compete in a licensing sense.
Re: (Score:2)
Yea, no shit. It's a bit like arguing that we don't need geometry, trigonometry and Calculus because natural number arithmetic is the oldest standard for mathematics.
Re: (Score:3)
I find open source to be useful even in cases where it isn't Free Software. There was a time, in the olden days, when you bought software libraries for development that were distributed as a binary and header. These binary only distributions sometimes required royalties, but it was also common to have been a one time charge and royalty free. Either way it sucked because you could not see how it worked, modified it for your needs, or ported it to a different platform. Then this idea of open source hit the sc
Re: (Score:2)
Re: (Score:1)
You're half right. Free software copyleft licenses privilege one type of freedom over another.
Whether you feel it as a restriction or a freedom depends on which part of the software ecosystem you find yourself in, and with which intentions.
Re: (Score:2)
Myself, I go for the open sauce solution. Sometimes new jars are way too hard to open, and you have to bang them against the counter to break the seal, etc.
Huh (Score:1)