[besson3c]

What is your take on this?

I've been thinking that it would be nice if support for both of these formats via HTML5 video in all major browsers became the norm, but I can't see this happening now. There is now some haggling over adding DRM to videos which is complicating this mini war between Google/Mozilla vs. Apple/Microsoft (what an odd alliance I can't remember where Opera stands in this...

What I don't want is a situation where developers have to pay somebody if they want to be able to embed HTML5 video in their apps (via Webkit or Gecko), or we get into this corporate war of control over what videos can play where, similar to the initial iTunes Music Store deployment, for the simple fact that this will probably be a nuisance to us regular customers at the very least.

I also don't want all of this to slow the complete adoption of the HTML5 spec.

[turtle777]

Originally Posted by besson3c 

I also don't want all of this to slow the complete adoption of the HTML5 spec.

You can thank Google for it.

They are the reason hat we will have to deal with f*cking stupid plugins in the future.

-t

[P]

WebM is increasingly looking like a long shot. My best guess is that it will turn into another VC-1. If you remember, MS intended VC-1 to be "almost as good as H.264, but with lower royalties". When they opened the spec, most people who contributed to H.264 put their hands up, claiming to have patents in it. In the end, VC-1 cost almost as much as H.264 and so died because if you had to pay anyway, why not go with the best?

Opera stands with WebM, but their market share is so tiny that they're basically irrelevant.

Google's change of stance doesn't really help WebM any - all it does is help Flash. It used to be that you could get WebM to work in any desktop browser with <video> support (with plugins in Safari and IE), and that's still the case. You need a Flash fallback for both H.264 and WebM, and you always did. What you can't do is make WebM work on a significant portion of mobiles. The only version that has it is Android from 2.3 and up, and since upgrades are the exception rather than the norm on that platform, it will take years until even a majority of Android phones can use it. With Nokia going Windows Mobile, Android won't be a behemoth the size of Windows on the desktop - it will probably be the biggest, but not dominating. Since Flash on Android is still too slow to be really useful, we're back to H.264 there for the next 2 years at least. Google's decision turned H.264 from inevitable at some point in the future to another also ran next to Flash.

The only good thing to come out of this is that Adobe finally got serious about upgrading Flash. Flash 10.2 is really quite decent as a video player.

[besson3c]

Originally Posted by turtle777 

You can thank Google for it.

They are the reason hat we will have to deal with f*cking stupid plugins in the future.

-t

I'm okay with what Google has done. Speed of adoption is nice, but there is also getting it right. There will probably be problems with the pay-to-play approach to embedding HTML5 video as the sole option.

[besson3c]

Originally Posted by P 

WebM is increasingly looking like a long shot. My best guess is that it will turn into another VC-1. If you remember, MS intended VC-1 to be "almost as good as H.264, but with lower royalties". When they opened the spec, most people who contributed to H.264 put their hands up, claiming to have patents in it. In the end, VC-1 cost almost as much as H.264 and so died because if you had to pay anyway, why not go with the best?

Opera stands with WebM, but their market share is so tiny that they're basically irrelevant.

Google's change of stance doesn't really help WebM any - all it does is help Flash. It used to be that you could get WebM to work in any desktop browser with <video> support (with plugins in Safari and IE), and that's still the case. You need a Flash fallback for both H.264 and WebM, and you always did. What you can't do is make WebM work on a significant portion of mobiles. The only version that has it is Android from 2.3 and up, and since upgrades are the exception rather than the norm on that platform, it will take years until even a majority of Android phones can use it. With Nokia going Windows Mobile, Android won't be a behemoth the size of Windows on the desktop - it will probably be the biggest, but not dominating. Since Flash on Android is still too slow to be really useful, we're back to H.264 there for the next 2 years at least. Google's decision turned H.264 from inevitable at some point in the future to another also ran next to Flash.

The only good thing to come out of this is that Adobe finally got serious about upgrading Flash. Flash 10.2 is really quite decent as a video player.

It will also take years before we can get rid of Flash fallbacks, so I'm okay with these sorts of "organic" delays. I just hope that corporate warfare will not create casualties out of us users (and developers).

[Cold Warrior]

Google isn't going to indemnify Android hardware makers, so if they support webm, they're open to lawsuits from MPEGLA. Seems like it'd be a better choice to pay for the best -- h264 -- and move beyond the uncertainty of patents and lawsuits.

If mobile OS and desktop OS makers are all supporting h264 (assuming Android manufacturers decide to), that takes care of the problem.

[Athens]

H.264 is the only thing I support.

It has full support and hardware support in Every modern Video card, iOS devices, Android Devices, WebOS Devices, Xbox, PS3, Car Decks, some DVD players and TVs and MP3 players.

ITs a mature codec and looks as good as DVD if encoded right. I have no desire for anything new. This debate means the best choice for developers will be to continue using Flash which is using H.264 for the future vs having a native single codec support built into the browser. If Developers have to encode 3 copies of video files to be compliant with all the formats they will just continue to use flash to make it easier.

H.264 is also the safest format to use because there are so many companies involved with the patent pool.

H.264 is supported by Microsoft and Apple, with IE and Safari so the 2 largest players for Operating systems natively support it. Its already bloody hard enough to get Apple and Microsoft to support the same standards as it is now.

The other issue is the quality of video on WebM. Its bad, I mean really bad. I been testing Youtube videos in WebM format and they are terrible. I also encoded 2 DVDs in WebM when this was first announced and compared it to my h.264 version. While a tiny bit smaller the quality of the WebM versions was just dreadful. Now I admit I could have messed something up during the encoding because it took a few attempts to get my DVDs looking great with h.264 but even my failed attempts with h.264 didn't look as bad as WebM.

If anything WebM is going to go the same path as PNG and Gif. By the time PNG was equal to GIF, GIF patents had run out making it pointless. WebM is so much further behind and video is so much more complicated, by the time we see WebM being equal to H.264 it might not matter any more.

I am very angry with Google for this and has made me question using any of their services in the future as a form of protest. Ive been testing out BING on my iphone, using Opera for browsing, and stopped using a few other google services already to test the waters of separation. I am about as deeply integrated into Googles ecosystem as it gets and im finding its not totally impossible to untether myself. Thought I might have to pay money for something like MobileME to maintain a good contact, email, calendar sync setup. In the end if I did do that and canceled my Gmail I wouldn't mind because MobileME is a better solution then Google Gmail/Conacts/Calendar when in a Mac and Windows environment. MobileME does better syncing with Outlook.

So we will see how this goes. The move to drop H.264 from Chrome which was and is my fav browser is a political move only. It is this aspect of it that makes me mad. Its one thing to add support in for WebM and to say hey this is what we want to move to. Its another thing to drop support and a feature out to further the above goal. Its also one thing to back something for a year and when every one is close to a final choice and it looks like we can move forward they drop the bomb at the last second resetting years of work to ground zero again.

Google has become dangerous and arrogant. And no other company on this planet has more data on us then Google. Facebook would be a distant second.

[Athens]

Originally Posted by Cold Warrior 

Google isn't going to indemnify Android hardware makers, so if they support webm, they're open to lawsuits from MPEGLA. Seems like it'd be a better choice to pay for the best -- h264 -- and move beyond the uncertainty of patents and lawsuits.

If mobile OS and desktop OS makers are all supporting h264 (assuming Android manufacturers decide to), that takes care of the problem.

Google already pays the full fee for H.264 with youtube so they had no extra costs involved with Android or Chrome. Dropping support from Chrome does not save them a penny either. This is more to do with politics unless Google really thinks they can drop H.264 from everything to save 14 million a year in royalties or whatever the current MPEGLA cap is for large companies. .

[Athens]

Mozilla are being idiots too because Microsoft offered to pay for the license for Firefox to support it natively so the HTML5 standard could be totally finished and we could all move on. Now I don't actually hate Mozilla for not accepting this because they made there position clear a long time ago and have been consistent. I think its a bone head mistake but they have a reason they have stuck with for a long time and respect that. Same goes for Opera. Neither have supported H.264 and never have had native support for it to take away.

[besson3c]

In the end if I did do that and canceled my Gmail I wouldn't mind because MobileME is a better solution then Google Gmail/Conacts/Calendar when in a Mac and Windows environment.

How do Windows users sync their calendars and contacts to MobileMe?

[besson3c]

Originally Posted by Athens 

Mozilla are being idiots too because Microsoft offered to pay for the license for Firefox to support it natively so the HTML5 standard could be totally finished and we could all move on.

The HTML5 standard is far from totally finished with or without video. Embedded media is only a small part of the HTML5 spec.

[P]

I don't have a problem with Mozilla here. Well, a little bit, but I can live with it. The Mozilla guys looked at <video> as a way to get rid of everything proprietary, and if it doesn't, they might as well skip <video> altogether. Apple's perspective was that it simply was a way to get rid of Flash and standardize the little workaround they had going with Youtube. The Mozilla guys probably felt that Apple came in from the left and stole their favorite tag. The problem I do have is the Mozilla's insistence on not using H.264 decoding if the ability is there in the OS. Their argument is basically that plugins are evil and need to die, and using code from the OS is just another plugin. This is bullshit (they use fonts from the OS, to name one) and inconsistent with their behavior otherwise, but the thing that really bothers me is that it puts the user experience second to their politics. This and a few other details (their sore loser reasoning around the Acid3 test) makes me think that they've grown cocky and lost sight of what's important here.

Google are the ones that I really have a problem with right now. They make their own standard and won't even submit it to be standardized - that's exactly what MS used to do in the nineties before everyone started calling them on it. They claim that there are no patents against it, but they themselves are already licensees to all the likely patents (those in H.264) and they won't indemnify anyone else. Not only do they they refuse to allow access to codecs in the OS, but they are the only party that have actively removed support for a video format from their browser. Unless they're getting paid by Adobe, the only possible reason is to mess with other mobile platforms. The really interesting bit is if they remove H.264 from Youtube. Let's see here, using a dominating position in one area (web video - Youtube) to expand into another (mobile OS) - isn't there a law against that?

[Spheric Harlot]

Originally Posted by Cold Warrior 

Google isn't going to indemnify Android hardware makers, so if they support webm, they're open to lawsuits from MPEGLA. Seems like it'd be a better choice to pay for the best -- h264 -- and move beyond the uncertainty of patents and lawsuits.

This right here is the crucial point.

As it stands, we have only Google's word that WebM is NOT patent-encumbered, and they're not willing to put their own money behind that claim.

Until that is finally cleared (and I find it difficult to believe that there won't be one or another patent that's violated somewhere in there), "royalty-free" is just a carrot made of bullshit.

All this move does is help Adobe and hurt Apple (as Flash remains a default player for another few years).

[mduell]

It's strange that the stage is set as H.264 vs WebM, not H.264 vs VP8. It's like comparing a car to an engine or something.

I think VP8 is a poor choice for HTML5 video for both academic and practical reasons: it was developed in secrecy without widespread interest, the "standard" uses code instead of words in some places, it's held by a single vendor instead of standardized by an independent agency, etc.

The compression quality/speed is a mixed bag. Based on the fundamentals it should be able to match H.264 Baseline Profile when a good encoder comes out (like xvp8).

MPEG-LA is working on a VP8 patent pool, so it will soon be as easy to license VP8 as is is H.264 or, if they don't find many/any patents to include, higher confidence of it's claimed royalty-free status.

I suspect the web will continue to be fragmented going forward. If there is a single standard universally supported, I think the most likely option is the royalty-free codec MPEG is talking about developing.

[P]

Originally Posted by mduell 

MPEG-LA is working on a VP8 patent pool, so it will soon be as easy to license VP8 as is is H.264 or, if they don't find many/any patents to include, higher confidence of it's claimed royalty-free status.

This is what I mean with the VC-1 situation. If VP8 (you're right that H.264 vs. VP8 is the right way to phrase it) isn't royalty free, why bother with it at all? Will Google buy out any patents MPEG-LA finds?

[lpkmckenna]

WebM is the stupidest Google scheme ever.

[OreoCookie]

IMHO Google's strategy will fail big time: let's forget the problems associated with the closed development history, poor documentation and coding as well as the licensing issues for the moment.

Even if you believe all the claims that Google makes, it is `open,' not encumbered by patents, etc., it just isn't a standard in the sense that there exists a standards body that allows other companies to build their, say, software and hardware decoders to specs. This means, there will be no hardware support in mobile SoCs for VP8 in the foreseeable future. This is the first big nail in the coffin.

The second one is that Google has chosen to keep Flash with built-in h.264 support as part of the browser! So if Google chose to remove h.264 support altogether (motivated by the philosophical argument that the `web should be based on open source technology' or something), their stance would be coherent, but instead, Google chooses a path that may make the <video> tag useless.

Even if I operate under the assumption that Google wants WebM to be ubiquitous, their strategy is a failure: instead of making it universally available first and then pulling the plug, they do it now when very, very few devices support it. Even most Android-based devices are incompatible (since the manufacturers do not offer updates to the latest version of Android).

[besson3c]

So what is the best strategy to help prevent pay-to-play, or is there one?

[P]

What do you mean by pay-to-play? The H.264 encoder and decoder costs money. Broadcasting on the Internet doesn't cost anything and never will.

[besson3c]

Originally Posted by P 

What do you mean by pay-to-play? The H.264 encoder and decoder costs money. Broadcasting on the Internet doesn't cost anything and never will.

It looks like it is royalty free for end-users to upload content (for now), but what about developers that want to, say, create a mobile app with a mini Webkit-based browser that plays HTML5 video?

[P]

No, not for now - for ever, that's the point. Any developer will pay though, but since there is H.264 decoders packed in with the OS in most cases, you can just use them.

[besson3c]

Originally Posted by P 

No, not for now - for ever, that's the point. Any developer will pay though, but since there is H.264 decoders packed in with the OS in most cases, you can just use them.

I thought you said that we'd have to pay for encoding and decoding? What if a website wanted to allow users to upload video of some format and have this converted to H.264 ala YouTube or something?

[mduell]

Originally Posted by P 

This is what I mean with the VC-1 situation. If VP8 (you're right that H.264 vs. VP8 is the right way to phrase it) isn't royalty free, why bother with it at all? Will Google buy out any patents MPEG-LA finds?

If there's fewer patents (and that seems likely, since a lot of the decisions made in VP8 appear to be avoiding patents), VP8 may be cheaper to license than H.264.

Originally Posted by besson3c 

So what is the best strategy to help prevent pay-to-play, or is there one?

Patent reform in the US, particularly with software patents.

Originally Posted by P 

What do you mean by pay-to-play? The H.264 encoder and decoder costs money. Broadcasting on the Internet doesn't cost anything and never will.

Broadcasting for free on the internet is free, but the conversion/production tools still need to be licensed, any paid site needs to be licensed, etc.

Originally Posted by OreoCookie 

Even if you believe all the claims that Google makes, it is `open,' not encumbered by patents, etc., it just isn't a standard in the sense that there exists a standards body that allows other companies to build their, say, software and hardware decoders to specs. This means, there will be no hardware support in mobile SoCs for VP8 in the foreseeable future. This is the first big nail in the coffin.

Most of the mobile DSP acceleration for video is generic building blocks (iDCT, iMDCT, motion comp, in loop deblocking) rather than specific "decode H.264 bitstreams."

Originally Posted by besson3c 

I thought you said that we'd have to pay for encoding and decoding? What if a website wanted to allow users to upload video of some format and have this converted to H.264 ala YouTube or something?

Converting to/from H.264 requires an MPEG-LA license. Some conversion tools (like say Apple's Compressor) come with one, but only for non-commercial use.

[besson3c]

How does the licensing/legal stuff differ between VP8 and H.264, exactly?

[mduell]

VP8: Dozens of known patents formerly owned by On2 now owned by Google under a royalty-free license. No other patents presently known/claimed publicly.

H.264: Hundreds of known patents in a pool (none known outside of it), requires a patent license to do just about anything (broadcast, convert, decode, etc), fees range from $0.01-0.20 per thing (subscriber, software/hardware shipped, video, whatever), all capped at $6.5M/yr.

[Athens]

Originally Posted by besson3c 

How does the licensing/legal stuff differ between VP8 and H.264, exactly?

If you are licensed with H.264 your safe. If you use VP8 and and its found to be violating patents you would be at risk for lawsuits to recover royalties you didn't pay for using a codec that was not patent free. Its a very high risk.

[P]

Originally Posted by mduell 

If there's fewer patents (and that seems likely, since a lot of the decisions made in VP8 appear to be avoiding patents), VP8 may be cheaper to license than H.264.

May be, but

1) The FSF et al don't care about price- 1 cent for the life of the program is enough. This is also the position for a standard to be included in a W3C specification. It's royalty free or bust.

2) That analysis of VP8 by an x264 developer was pretty damning - VP8 is uncomfortably close to H.264.

3) If I have 7 patents in H.264 and 1 in VP8 and VP8 is set up to compete with H.264, it makes sense for me to jack up the price for that single patent.

The only way this works is if Google buys all the patents that are missing.

Originally Posted by mduell 

Most of the mobile DSP acceleration for video is generic building blocks (iDCT, iMDCT, motion comp, in loop deblocking) rather than specific "decode H.264 bitstreams."

All those things are indeed generic, but the most processor intense bit is the entropy coding. VP8 supports an entropy coding similar to, but not identical to, H.264 CABAC. Either mobile DSPs need to support this, or VP8 will have to skip its entropy coding.

(To be clear: H.264 Baseline profile, which is used in mobiles, does not use CABAC. That's fine, but a) VP8 does not seem to have the same profile system and b) VP8 without its entropy coding is a much weaker codec and not really comparable to H.264)

[OreoCookie]

Originally Posted by mduell 

Most of the mobile DSP acceleration for video is generic building blocks (iDCT, iMDCT, motion comp, in loop deblocking) rather than specific "decode H.264 bitstreams."

Sure, dedicated hardware is useless without proper software -- that's exactly my point. The problem of having VP8 hardware de- and encoding support has less to do with technological limitations, but rather with the lack of industry support. Why bother right now? Are there any major video platforms that require you to use VP8 instead of h.264? I'm not aware of any. To my knowledge, there is no support amongst professional video software either.

[Athens]

And if you look at history, we went through this with GIF and PNG a decade ago. The patent holder of the compression used in GIF wanted money for the use of the format so the community started work on PNG. By the time PNG was good enough to replace GIF the patents had ran out for GIF making it useless from the point of why it was developed in the first place. Today PNG is better offering better compression, more features and true color and its still hardly used compared to GIF. APNG which is only supported in Firefox brings animation to PNG and I have yet to see a single APNG file ever in my life and its been out since 2004. By the time the patent ran out for GIF PNG was a mature format so today we see its use along side GIF. Video is way more complex. And add to that it could take a decade for mass adoption, hardware support, the legal status to be all sorted out and by then the patents for H.264 might not be a issue any more. Ogg Vorbis was created to avoid MP3 patents and to this day is still a very limited supported format. H.264 patents mostly expire in 2025, thats 14 years from now. Between the legal battles that will occur not if but will occur over V8, getting hardware support, consumer support, vender support and professionals supporting the format I expect the h.264 patents will have expired before V8 even reaches 25% of the market in use. So I see no point!!!! Learn from history. Learn from the PNG and GIF example and the Ogg and MP3 example.

[Athens]

BTW more facts about h.264

Some of the major players in the patent pool are Microsoft, Apple, Sony, Dolby, Toshiba and Panasonic. You can bet Sony, Toshiba and Panasonic will never provide hardware support in Cameras for V8, and a lot of the home and professional cameras are from those companies. Now a lot of people are saying that Apple and Micorsoft are evil because they belong to this patent pool. Apple has a single patent involved with H.264 and Microsoft has 75. Currently Microsoft is paying MORE in license fee's then its getting back in royalities. So neither have reason to support this considering they both lose money as they both have to license H.264. So why do they support it then if it costs them money. Well it has to do with having a single standard which makes things a lot easier for all. Otherwise Betamax would be around with VHS, and HD DVD would still be around with Blu-Ray and we would all be happy in our fragmented universe.

The other thing that V8 needs to watch for is patents owned by AT&T which isnt part of the MPEG-LA patent pool. This means even if V8 gets all the stuff sorted out with MPEG-LA they still could be subject to being sued by AT&T. AT&T has a cross license agreement with MPEG-LA so your protected from AT&T with h.264 as well.

The basic fees involved with H.264 are from the encoder and those selling or making profit off of providing h.264 movies. Not the end user watching it or people making blog videos with no commerical value. Another key issue with h.264 is that its free to use for free internet videos. Will not cost anything for distributor or end user for internet video.

[turtle777]

What Athens writes makes sense.

I don't see how V8 will be widely adopted anytime soon, and at some point, the soon-to-be patent free h.264 will look even better.

One thing that goes along with this: as bandwidth becomes more and more, and readily available, there will be less and less a need for more optimized codecs for video. Basically, like with GIF, saving a few byte becaomse a moot point.
So will it be with h.264. The current iteration is good (enough), and will stick around for a long time.

-t

[mduell]

Originally Posted by P 

2) That analysis of VP8 by an x264 developer was pretty damning - VP8 is uncomfortably close to H.264.

JGG's blog post also notes that many of the differences between VP8 and H.264 are to avoid H.264 patents.

Originally Posted by OreoCookie 

Sure, dedicated hardware is useless without proper software -- that's exactly my point. The problem of having VP8 hardware de- and encoding support has less to do with technological limitations, but rather with the lack of industry support.

My point was mobile devices don't need new hardware to accelerate VP8.

Originally Posted by turtle777 

I don't see how V8 will be widely adopted anytime soon, and at some point, the soon-to-be patent free h.264 will look even better.

Before H.264 is patent free in 14 years, I suspect we'll all move on to H.265.

Originally Posted by turtle777 

One thing that goes along with this: as bandwidth becomes more and more, and readily available, there will be less and less a need for more optimized codecs for video.

Video resolution is going up, framerate is going up, 3D is being pushed: I suspect there will continue to be pressure on codec optimization.

[Athens]

next step will be providing high def content at a reasonable size. 2-3 GB per movie is still a lot, and thats at 720p, pressure will be on for 1080p in the future too at a reasonable size.

[OreoCookie]

@Athens
And in many respects, png is technologically superior to gif, was soon supported by all major browsers and image editing software -- something that cannot be claimed for VP8! But even then, png hasn't replaced the gif format.

[Athens]

In all respects its superior to GIF now with the exception of animation which APNG does. It can do true color, at 8 bit color it has better compression then GIF for smaller files. Has other features which makes it much superior to GIF. But support was lacking back in the days when GIF still had patents on it. By the time Browsers started to support PNG it didn't matter because the patents for GIF ran out. It was not until IE 7 that PNG had native support in IE years after it didn't matter. I see the exact same pattern with VP8. By the time its equal or better to h.264 it wont matter.