Hulu Joins Netflix and Amazon In Promoting Royalty-free Video Codec AV1 (fiercecable.com) 134
theweatherelectric writes: Hulu has joined the Alliance for Open Media, which is developing an open, royalty-free video format called AV1. AV1 is targeting better performance than H.265 and, unlike H.265, will be licensed under royalty-free terms for all use cases. The top three over-the-top SVOD services (Netflix, Amazon, and Hulu) are now all members of the alliance. In joining the alliance, Hulu hopes "to accelerate development and facilitate friction-free adoption of new media technologies that benefit the streaming media industry and [its] viewers."
Encoding AV1 is computationally expensive (Score:2, Interesting)
Currently AV1 encoding with common encoding tools is a very time consuming process, as can be seen in the below screenshot taken from a Lenovo T540p notebook with an i7-4800MQ, 8GB RAM running Ubuntu 14.04. It would take 8 hours and 42 minutes to encode a 1080p@24fps 40 second long sequence (Tears of Steel Teaser) with a target bitrate of 1.5Mbps.
I wonder if GPUs can speed things up?
Re:Encoding AV1 is computationally expensive (Score:4, Informative)
AV1 still alpha (Score:4, Insightful)
That, and writing a non-prototype encoder, most likely.
yup, currently AV-1 is still an alpha.
it's still a playground in which to experiment by activating feature which are currently being developped.
(e.g.: the Perceptual Vector Quantization (PVQ) and Assymetic Numeric System entropy coder (ANS) that were developped at Xiph as part of Daala, can be tested into AV-1)
Wait until it hits AV-1, only then will developers start optimizing performance instead of chasing compression factors.
Re: (Score:2)
Re:Encoding AV1 is computationally expensive (Score:4, Insightful)
For home use, I don't really see the point of using these very computationally expensive codecs - it's not like you can make better rips... just smaller ones, and disk space isn't expensive anymore. My hundred-or-so DVD/Blu-Ray collection was ripped to h.264 a number of years ago, and those still work just fine.
However for a commercial service, it's a different argument. Not only do they have tens of thousands of items in their catalogs, but there's also bandwidth to think about. For them, the investment may make sense. However if it's equally expensive, hardware-wise, to decode the streams... then they have to worry whether their customers will be willing to make the investment.
Re: (Score:2)
No worries dude. All these codecs are designed to be easier to decode them than it is to decode them, so inexpensive-real-time-decoding players for media consumption are feasible on day one (say, either streaming from the internet, or streaming from a plastic shiny disk spinning at a fixed rate).
Normally you would produce content in non-compressed format for maximum quality, and then compress in non real-time. As time progresses, and moore's law progresses (and the programmers codding the codec refine the a
Re: (Score:2)
[...] designed to be easier to decode them than it is to decode them, [...]
lulwut?
Re: (Score:1)
I'd like to see you watch anything using HVEC (265) codec. It has to be the worst out there and I wish it would go away.
Re:Encoding AV1 is computationally expensive (Score:4, Insightful)
My hundred-or-so DVD/Blu-Ray collection was ripped to h.264 a number of years ago, and those still work just fine.
So you compressed a high quality source into a smaller file, but you say there's no point in potentially doing it with better quality? You still have the original collection then you could get a quality improvement.
If you don't have the original however you should note that files aren't getting smaller, and the "not expensive" 4TB HDD will quickly fill up if you value your 4K content.
Re: Encoding AV1 is computationally expensive (Score:1)
Re: Encoding AV1 is computationally expensive (Score:4, Informative)
Quality of MPEG-2 is always greater than MPEG-2 re-encoded into mp4.
There is some encoding and filter tricks you can do to hide the loss of quality somewhat, but is still a loss.
Re: (Score:1)
Well when I'm wanking off to pr0n that has been ripped from DVD/MPEG-2 and re-encoded, I always notice the loss of pleasure. I attribute it to generational loss. I honestly prefer to wank directly from the DVD. And not only is the pleaure heightend, but I also notice a fuller more abundant wad of semen.
Re: (Score:3)
Since you ignored bitrate.
Since you ignored that a lossy source is always higher quality than a second lossy recompression of the original lossy source.
Re: (Score:2)
For home use, I don't really see the point of using these very computationally expensive codecs - it's not like you can make better rips... just smaller ones, and disk space isn't expensive anymore. My hundred-or-so DVD/Blu-Ray collection was ripped to h.264 a number of years ago, and those still work just fine.
However for a commercial service, it's a different argument. Not only do they have tens of thousands of items in their catalogs, but there's also bandwidth to think about. For them, the investment may make sense. However if it's equally expensive, hardware-wise, to decode the streams... then they have to worry whether their customers will be willing to make the investment.
Don't see the point?? Who buys DVDs snd Blu-Rays these days to rip them? I never even bothered to get a Blu-Ray player. Netflix, Amazon, and Hulu are streaming services. They presumably want to use this codec for streaming, in which case chewing up computational resources which are available in plenty on most PCs, is an acceptable tradeoff for better quality and above all probably less bandwidth consumption. This is especially important now that Trump's new FCC Chairman Ajit Pai is getting ready to stomp th
Re: Encoding AV1 is computationally expensive (Score:2)
Re: (Score:2)
With hardware acceleration AV1 is not that battery intensive. Just about everyone is adopting AV1 except Apple. What makes a big difference is if the underlying hardware supports offloading AV1 encoding and decoding. All of the major hardware vendors will support AV1, including Intel, AMD, ARM and Nvidia. Apple is the lone holdout. Most web browsers except Safari support AV1 which builds on VP9. VP9 is also not supported by Apple.
Re: (Score:1)
The keyword here is royalty free... h.264 is encumbered by MPEG-LA
Re: (Score:2)
Later hardware support will appear both for PC GPUs and for mobile SOCs (there're hardware manufacturers in the alliance) and it'll be feasible to play even on mobile devices.
Meanwhile it can coexist with current codecs.
Funnyly, this is one of those cases in which a few companies (alliance for open media) ally to fight other companies (h265's paten
Re: (Score:1)
a Lenovo T540p notebook with an i7-4800MQ, 8GB RAM
There's your problem. You have to use the proper tool for the job.
8 hours and 42 minutes to encode a 1080p@24fps 40 second long sequence with a target bitrate of 1.5Mbps.
For comparison, I'm using a PC with 32GB RAM and an AMD 8 core CPU (the old FX, not the new Ryzen)
It takes approx. 6 hours to encode a 2 hour 30 minute video, 1080p@25fps, HEVC/H.265, with a bitrate of 5Mbps.
Re: Encoding AV1 is computationally expensive (Score:1)
Re: (Score:1)
GPU : Yes (Score:5, Informative)
I wonder if GPUs can speed things up?
given that [aomedia.org] AMD, Nvidia, Intel, ARM, Broadcom are also on board (beside content providers like Netflix, Amazon, Hulu and Google)
you can bet that Yes, there are going to be GPU implementations.
(And if you've followed the posts of Xiph - you know that they take GPU into account from the beginning).
Also there are already currently cloud based solution [bitmovin.com] that distribute the compression workload accross a cluster.
(Video is split into smaller segment, each segment is independently compressed by a separate job on the cluster, then the compressed streams are concatenated together).
And bitmovin is already providing alpha support for AV-1 as it is now (so they can already test their solution and so, in a few months, on the day when AV-1 hits version 1.00 they are already ready and their users have already tested pipelines).
Actually the only single major player that is missing here is Apple.
Probably because they are betting all their marbles on their own patended H265/MPEG4 HEVC.
They are among the patent owners of the patent - so using/licensing H265 comes much more cheaper for them.
Which was the main reason for everybody else to drop H265 and consider joining Aomedia for AV-1 (between the original patent-pool, the other competing pools that have formed with other sets of patents and patent troll waiting to sue to try to get their share, licensing H265 is a much more expensive adventure than licensing H264/MPEG4 AVC was- To the point that H265 licenses cost a significant part of the price of embed ARM SoC as those used by cheap phones, ruining their competitivity)
Linux : Yes. (Score:2)
This should come to Kodi, right?
Right !
ARM and Broadcom are on the AOMedia.
So if you use some tiny ARM computer board you should be covered (by the time Raspberry Pi 4 or 5 is out, its video-core should be able to handle AV-1).
I use Linux on my home-made quad core desktop.
AMD, Nvidia and Intel are also on board.
So probably your future Radeon or Nvidia GPU is going to be able to handle it too.
Re: (Score:2)
Every single Pi uses the same GPU with the same capabilities (other than some clockspeed changes). They don't appear to have any plans to change it, and the next revision isn't expected until 2019 or later, so I wouldn't expect AV-1 support any time soon.
AV1 is interesting (Score:2, Funny)
but I'm going to wait for AV2
Re: (Score:2)
Re: (Score:2)
Followed by AV0 - the prequel.
Re: (Score:2)
Then just 'AV', the reboot.
Re: (Score:2)
Then the crossover, AV v. VC1.
Re: AV1 is interesting (Score:2)
Re: (Score:1)
You should wait for AV3.11 for workgroups.
Re: AV1 is interesting (Score:2)
Re:Oh... (Score:4, Insightful)
Re: (Score:1)
>Even an evil clock is just twice a day
No, that's not how evil.
While nice... (Score:2, Interesting)
It will take about 10 years for it to become a viable standard. Considering how many devices out there that won't support it. I know that I won't rush out to replace my Smart TV that can now handle H265 and H264. Nor will I be re-encoding my videos until forced to, which would be around 15-20 years IF IF IF this 'new video code' (heard this story before) becomes viable.
Re: While nice... (Score:1)
Your argument could equally be said about displayport yet hmdi won out on tvs, consoles, av receivers, settop boxes, pvrs.
One big reason is displayport was slow to add ethernet, audio and drm to the spec but this is the same issue with av1. It is in alpha stage while h.265 is ready now with nvidia and amd already shipping cards long ago with h.265 hardware encoding and there are dozens of sub $100 arm set top boxes which do h.265 hardware decoding.
Re: (Score:2)
It will take about 10 years for it to become a viable standard.
No, the AV1 bitstream format will be frozen later this year, browsers will add support for AV1 soon after that (Mozilla, for example, is already working on it [mozilla.org] in Firefox), and YouTube, which is the world's largest video site, intends to start using AV1 as soon as possible. AV1 will be adopted quickly.
Considering how many devices out there that won't support it.
Many devices will be able to support it in software. My iPhone 7 doesn't "officially" support VP9 [wikipedia.org] but VP9 video plays back just fine in VLC for iOS.
Nor will I be re-encoding my videos until forced
You don't have to re-encode anything unless you want to.
Re: (Score:3)
Hardware VP9 decoding is rare enough of a feature already.
It's not that rare. Intel's been shipping VP9 decode acceleration [techreport.com] for two years now. Android has supported VP9 decoding [android.com] since Android 4.4, which was released in late 2013. If you have an Android phone, you probably have VP9 hardware acceleration. Plenty of AV1 hardware will be released in late 2018.
But also don't underestimate today's mobile devices. I have an iPhone 7 and I can play VP9 video in software in VLC for iOS [videolan.org] without issue. A future VLC update will add AV1 support.
Re: (Score:2)
I have an iPhone 7 and I can play VP9 video in software in VLC for iOS without issue.
Is shorter battery runtime not an "issue"? Or not being able to watch videos embedded in the webpage, especially videos that use Media Source Extensions or Encrypted Media Extensions?
Re: (Score:2)
Is shorter battery runtime not an "issue"?
No.
Or not being able to watch videos embedded in the webpage, especially videos that use Media Source Extensions or Encrypted Media Extensions?
Also no. Safari doesn't support VP9 so it won't be served up to Safari.
Re: (Score:2)
Safari doesn't support VP9 so it won't be served up to Safari.
Instead, a notice like the following would be served to Safari, Chrome for iOS, Firefox for iOS, and every other web browser for iOS.
Would you find it acceptable if a growing fraction of web videos started displaying notices like this?
Re: (Score:2)
Instead, a notice like the following would be served to Safari, Chrome for iOS, Firefox for iOS, and every other web browser for iOS.
No, it will just serve H.264. All video streaming services will encode to H.264 as their baseline, fallback format and offer better formats as an option, which is exactly what they do now. If you can't play AV1 or VP9 you will be served H.264. That has been and will be the status quo for years to come.
Royalty for H.264 on your personal MediaGoblin (Score:2)
No, it will just serve H.264. All video streaming services will encode to H.264 as their baseline
Good for incumbents like Dailymotion, Vimeo, and YouTube. But how would the operator of a new video streaming service, such as a hobbyist operating a video streaming service on his own domain to stream his own videos, afford the patent royalty for use of each encoder used to transcode video to H.264 and audio to AAC?
Re: (Score:2)
But how would the operator of a new video streaming service
Who cares? That's their problem to solve. Don't forget all the investment in hardware and services just to start this streaming service in the first place. How are they going to afford that?
No one says they have to support iOS, just as no one says they can't go ahead and release an app for iOS which decodes whatever video format they want to use.
Re: (Score:2)
But how would the operator of a new video streaming service
Who cares? That's their problem to solve.
For example, if theweatherelectric were to produce video and exhibit it to the public, it would be theweatherelectric's problem to solve. In such a situation, how would theweatherelectric afford the H.264 encoder royalties?
Don't forget all the investment in hardware and services just to start this streaming service in the first place.
How much does a VPS capable of running MediaGoblin cost lately?
Re: (Score:2)
How much does a VPS capable of running MediaGoblin cost lately?
You tell me. This is your directionless hypothetical scenario so you answer the question. And what about lunch? Do they have lunch money? How will they afford lunch?
Re: Royalty for H.264 on your personal MediaGoblin (Score:2)
Asking absurd hypotheticals is what tepples do best.
Yeah? Well, what if they wanted to do this on hardware from 2008?
It's kinda what he does, pretty much always. Meh... Sometimes they are legit questions.
Re: (Score:2)
Lets see the forces opposed to your choice. All media companies distributing compressed digital content (https://en.wikipedia.org/wiki/High_Efficiency_Video_Coding pretty fucking expensive 25 million dollars, for something that is a straight mathematical model pretty much any company could have come up with, nothing new, just application maths). Manufactures also save the fee and as a major bonus get to replace all outdated equipment or if the customer base in too cranky, supply a software update.
So billio
Re: (Score:1)
My devices that can handle streaming video include a TiVo, PS3, PS4, and an Amazon fire stick. All of them are software upgradable. A lot of smart TVs are Android based, so they are as well. A company like Netflix has decent bandwidth savings if only a few percent of its users switch, and the switchover will be nearly invisible to them. The app will just query the hardware to find what codec is available.
Obviously there's no
Re: (Score:2)
Well, it's not like I like Apple's 4th-generation Apple TV anyway. Stupid "touch interface everywhere!" mentality. When my 3rd-generation Apple TV isn't compatible with Netflix anymore, I'll switch to something else.
Which box is the best for Netflix? Take note that I'm in Canada, so I don't care about Amazon/Hulu/whatever USA-only-streaming-services.
Sheesh. Welcome to the party, pal. (Score:2, Insightful)
What about Dirac? Invented for the exact same reason. Theora anyone? Same thing. VP1? Again.
What's got me slightly pissed off is why the fuck these assholes all went "Nope, fuck off" to all of those in turn? Were they hoping to make enough money with locked down codecs at the time that they wanted the ability to enforce rights in codecs? Or just NIH?
Re: (Score:3, Insightful)
It takes a long time to go from inventing the standard and producing a sufficiently competent encoder. Hell, look at mp3 encoding and how right now lame is tons better than the first mp3 encoder. Yet...
Re: (Score:1)
Or maybe you don't know what you're talking about?
AV1 is an extension of VP9, which traces back to VP1, and includes Ophus for audio, developed by Xiph, as a succesor to Vorbis and Speex. Dirac is actually the only NIH project developed by a single "company": the BBC.
Re: (Score:3, Insightful)
The economics changed. There are enough large streaming services now that it's cheaper for each of them to work with the others developing a new format than to keep licensing the latest and greatest codecs from MPEG LA. Older codecs aren't up to the challenge of streaming 4K video over the shitty connections that pass for broadband in the States.
Re: (Score:2)
Did you check who are involved ? These are the same people who worked on VP and Dirac and then some companies that know how to do streaming.
Mostly the same core companies that were working on Opus at IETF, they started this work at the IETF as well. Not sure why they went their separate way for this though.
Re: (Score:3)
Not sure why they went their separate way for this though.
AOMedia and NetVC [ietf.org] are not separate efforts, they complement each other [mozilla.org]. The same people involved with NetVC work on AV1. NetVC will use AV1.
Re:Sheesh. Welcome to the party, pal. (Score:5, Informative)
What's got me slightly pissed off is why the fuck these assholes all went "Nope, fuck off" to all of those in turn?
They didn't. VP9 [wikipedia.org] is used, for example, by YouTube [googleblog.com], Netflix [medium.com], and Wikipedia. Watch a video on YouTube, right click on it and select "Stats for nerds". If your browser supports VP9 then chances are the video will be playing back in VP9.
AV1 is the successor to VP9.
Re: (Score:2)
Could this benefit physical media advancement? (Score:1)
Re: (Score:2)
VHS tapes are reaching the point where they are degrading. You can still rip Blu-Ray and DVD.
Re: (Score:3)
But even then they are salvageable, even if a big chunk is damaged. Some media files won't play correctly if even 10% is damaged.
Big scratch on a DVD disc on the wrong side of the disc? might render it unplayable. VHS that got eaten by a deck? simple splice job...
(thinking about delamination problems on laserdiscs, even worse)
Re: (Score:2)
Which is why DRM is such a bad thing...
Digital brings the benefit of perfect copies, you can backup the media and keep the original safe. A spliced VHS tape may be playable while damaged, but a digital backup would be perfect.
When dealing with kids, or media that will be played/kept in hostile conditions, it's always sensible to make backups.
Why don't they let us know what encoders were used (Score:4, Informative)
Re: (Score:2)
Re: (Score:2)
This matters tremendously.
Indeed. Especially if you are using GPU acceleration as well. But I wonder if that is part of the equation. NVENC produces garbage comparable to FFMPEG's H.265 encoder, but I still use it because a 10x speed increase matters sometimes. I imagine it matters even more when you're serving half the internet's bits to customers.
silicon still matters (Score:1)
Re:silicon still matters (Score:4, Informative)
Re: (Score:1)
They are all chomping at the bit to have everybody all buy new hardware because the hardware they already have won't run this new thing...
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Are you joking?
Microsoft pays about twice as much in H.264 licensing fees [microsoft.com] as they receive in licensing payments. So in effect their use of H.264 is discounted by about 50%, but they're not making a profit out of it.
Re: (Score:2)
Microsoft receives back from MPEG-LA less than half the amount for the patent rights that it contributes because there are many other companies that provide the licensed functionality in content and products that sell in high volume.
So MS pays money for each installation of Windows; however, MS receives less money for content. That seems to be a problem for them as other companies sell vastly more content. For example iTunes probably sells way, way more content than Windows store. Amazon does too.
Money decision (Score:1)
This has nothing to do with any of this companies standing up for open access or freedoms. It is all about not wanting to pay the royalties and keep all the money for themselves.
Re: (Score:3)
It is all about not wanting to pay the royalties
Yes. I, too, don't want to pay royalties just to work with a video file and transmit it over the internet. I don't see video licensing as needing to be different from and I want it to be the same as HTML and PNG and JPEG and all the royalty-free protocols and formats that make the web and the internet possible. We have that now with VP9 [wikipedia.org] and we will have it with AV1.
In this instance their agenda matches our agenda. That's a good thing. Exploit it.
Re: (Score:2)
But hopefully AV1 can be software decoded on slow old hardware in in the 240p to 360p range and hopefully it actually gets used in such cases along with Opus.
1080p VP9 video works fine on old hardware. I've tried 1080p VP9 video on a desktop system from 2006 and worked fine playing YouTube in Firefox (it couldn't handle 4K video though). AV1 might end up being similar.
So do I have this right? (Score:2)
So do I have this right? Basically AV1 is VP10, but with two pieces taken from Daala (new symbol coding inspired by Daala, and the directional deringer)?
I'm not sure what other pieces ended up in AV1. There must have been something from Thor in there.
Re: So do I have this right? (Score:2, Informative)
Opus is also in there as the audio portion.
So can we retire gif now (Score:1)
Re: (Score:2)
What's a better term for "low-definition silent video, often with a low and/or variable frame rate, no longer than 15 seconds"?
Re: (Score:2)
Anything you can do with animated GIF you can do better with H.264 or similar
I'll believe that once you address these in a useful manner:
A. Browsers expect MP4 container to be in a <video> tag, not an <img> tag. Several websites allow users to add posts containing <img> but not <video>.
B. The last GIF-related patent expired in 2004. H.264 is still patented.
C. Does H.264 in MP4 container support variable frame rate? GIF frame rates are in units of 10 ms. Could this be worked around by encoding at 100 fps and using repeat frames?
D. Does H.264 in MP4 container e
Irony (Score:4, Interesting)
There's something very ironic about these three companies joining an "open media" alliance, while they all rely on DRM *extensively*.
Re:go away (Score:5, Funny)
Re: go away (Score:2)
Re: (Score:2)
Encoding movies with RealMedia is fun, but it's even more fun using it to post to Sla[BUFFERING]
Re: (Score:1)
You shouldn't need multicore processors to play a video file. That's why 265 sucks.
Re:Still image encoder (Score:4, Interesting)
Like Apple pushing for the HEIF that uses the h265 codec.
During their HEIF presentation at WWDC 2017 (video and transcript [apple.com], slides [apple.com]) Apple made the point that the HEIF format is designed to be codec agnostic. Apple will be using HEIF with H.264 and H.265, but in principle you could use any codec inside HEIF. HEIF [wikipedia.org] itself is just an image container format.
I imagine Apple will support AV1 eventually. If and when they do, they could go ahead and use AV1 in HEIF.
Re: (Score:2)
https://nokiatech.github.io/heif/ which is presented as the official site.
It's better to look directly at the HEIF git repository [github.com] than the website. To quote from the HEIF README: "HEIF is a media container format. It is not an image or video encoder per se. Hence, the quality of the visual media depends highly on the proper usage of visual media encoder (e.g. HEVC). Current standard allows containing HEVC/AVC/JPEG encoded bitstreams. This can be easily extended to future visual media codecs."
So right now HEIF supports AVC (H.264), HEVC (H.265), and JPEG. And in the future in ca