10 Questions for John Gruber Regarding H.264, WebM
osnews.com
osnews.com
Second, virtually all of these questions take as an axiom that H.264 is IP-encumbered and WebM/VP8 isn't. Technical analyses (by people who have actually implemented H.264 and are familiar with the patents) suggests that that simply isn't the case. That the author of this article thinks that VP3's release prior to H.264 might invalidate any part of the H.264 patent pool suggests that he may not be at all familiar with the patents in play.
Similarly, support for a "known patent troll" isn't germane to the debate if that patent troll is in the mix one way or the other, which appears to be the case.
Third, the fact that the H.264 patent pool hasn't been brought to bear on On2 tells us nothing at all about how successful H.264 would be against WebM; it's not at all unlikely that MPEG-LA sees nothing worth suing in Xiph/On2, and if that's the case, it is surely a different situation once Google pushes adoption.
Fourth, an "open pledge of support" for WebM from chip manufacturers is worth exactly nothing to the people who have spent many many hundreds of millions of dollars on mobile devices with hardware-accelerated H.264. One can reasonably argue that those people don't matter, but they can't with a straight face suggest that a "pledge of support" in any way mitigates the problem.
There's nothing wrong with this post as a contribution to the WebM/H.264 Apple/Google debate. But one gets the sense that the author sees it as some sort of devastating argument. I'm not sure he realizes that handwaving around "known patent trolls" and "decades of threats by MPEG-LA" and "is it because said codec isn't promoted by Apple", he's actually making a fairly weak sounding argument. He sounds emotional.
I'm gonna go with the analysis of the guy who said his B-frame implementation gave x264 the project's single greatest encoding quality improvement over the guy who thinks the release date of VP3 determines which patents encumber VP8, thanks.
The idea that VP3's prior release making AVC infringing is pretty stupid.
I'm not expert on video patents either. The overarching issue is, this post seems to think the IP situation with WebM is virtually cut-and-dry, and we have every reason to believe it isn't.
I see this a lot. It purports that Jason Garrett-Glaser has done an expert analysis of all of MPEG-LAs patents and has proclaimed specific infringement by VP8.
Of course that's complete bullshit. His analysis was superficially that there are similarities, and thus there may be infringement. Not one actual patent claim was shown to be infringed upon, much less even referenced to show that there was any understanding of it.
His analysis was as a codec guy. Not as a patent guy.
The devil are in the details.
It turns out (ha ha!) that when people vying for standards lay their patent traps, they tend to be extremely specific, ensuring that they don't get caught by prior art (there is plenty), but that they trap anyone who implements the standard. See Microsoft ActiveSync patents as a perfect example of this.
Does WebM infringe patents? Almost without question, as is the nature of the ridiculously generous patent system. Just as h264 is almost certainly infringing on patents of other parties that just haven't come forward to make their claims yet (and for which a h264 license is zero protection).
But one gets the sense that the author sees it as some sort of devastating argument.
Just as Gruber's post was framed. Google supports Flash, therefore they can't possibly be open, therefore Google is closed, therefore Google is evil...or something like that. Preaching to the choir and all that so I don't think much critical thought is necessary.
On the off chance that it's helpful, here's the author's bio:
My name is Thom Arvid Holwerda. I was born on the 1st of December in 1984, at the Medical Centre in Alkmaar, Noord-Holland, The Netherlands. Currently I live in Warmenhuizen, a small village a few miles north of Alkmaar (about 45 minutes by car north of Amsterdam). I've finished Latin/Greek school, and studied Psychology for two years at the Vrije Universiteit in Amsterdam (birthplace of MINIX); however, Psychology was not my thing, so I switched to something else. I'm now studying Information & Communication Sciences; a study which, other than the name might suggest, focuses mainly on language. About 50% of the study consists of studying a modern (to me) foreign language (Spanish, French, German, Italian or, in my case, English). My master will be Journalism.
You're right, Garret-Glaser isn't a patent attorney. On the other hand, he's an x264 developer intimately familiar with the standard.
In other words, he's definitely qualified to comment how similar VP8 might be, which few people deny, and not very well qualified to comment how infringing VP8 might be.
"MPEG LA doesn't favor one codec technology over another; we are like a convenience store that offers patent licenses for any number of codecs as a service to the market." (quoted in "Patent cloud looms over Google Web video plan," http://news.cnet.com/8301-30685_3-20006245-264.html?tag=mnco...)
Infringing or not, you may be like the local merchant who is forced to buy "insurance" to stay safe.
WebM is arguably the currently biggest threat to the goose that lays MPEG-LA's golden eggs. See also how they backtracked on royalties regarding web usage (even though you're still inviting a trojan codec into your home).
It reminds me at Microsoft's 250 or so patents over Linux, "but we won't tell you what they are, sucker"!
Put up, MPEG-LA, or shut up!
Because if we did see action, we'd find out once for and for all whether their innuendo has anything solid behind it. It's the standard patent FUD process, aided by everyone who says "it's probably patent encumbered", "we don't know whether it's unencumbered".
It's much better to imply they can crush you rather than actually try to crush you and risk failure. And the best part is that by implying they can crush you but not actually doing so, they can appear to be exercising benevolent goodwill while the FUD does most of the crushing for them, risk-free.
The impression I get is that everyone who brings up the "but MPEG-LA hasn't sued anyone, just threatened, they're all talk!" objection also believes that lawsuits are straightforward and that any company with a couple mil in the bank can launch one at a moment's notice. It's true, companies can sue you at a moment's notice, because it doesn't much to sue you and there's little to lose. The situation is a lot different when it's MPEG-LA vs. a patent infringer plus the resources of everyone who would like to see the patent portfolio fail.
Interestingly, while the Flash IDE is most definitely closed, the SWF file format and the tools to create SWFs... it's not so cut and dry. Here's the (Flash 10 SWF Spec)[http://www.adobe.com/content/dam/Adobe/en/devnet/swf/pdf/swf...]
There's probably a bunch of stuff left out that'd make implementation difficult, but that hasn't stopped:
(GORDON)[https://github.com/tobeytailor/gordon] -- an open source Flash runtime written in pure JavaScript . It only supports up to Flash v. 2; almost useless, but interesting.
Here's the (Flex SDK)[http://opensource.adobe.com/wiki/display/flexsdk/Flex+SDK] which includes the Flex framework, and a partially open source SWF compiler.
Apparently, Flex compiler isn't totally open source because of patents concerning audio and video codecs (I'm recalling this off the top of my head; please verify).
Here's (Tamarin)[http://www.mozilla.org/projects/tamarin/]. It's a VM for EcmaScript. It's been rolled into SpiderMonkey, the JS engine for FireFox, I think. It includes support for packages, namespaces, classes, and optional strict typing of variables. Really useful stuff for interactive devs that, unfortunately, hasn't been widely adopted.
Adobe contributes to WebKit on behalf of Flash & Air; I don't know what their contributions have been/how useful they are, though.
I'm a professional interactive dev, and I specialize in Flash, so you might conclude that I'd be biased, and I wouldn't blame you at all. However, as a Flash dev, I'm familiar with the non-marketed vectors of the Flash environment. Take that as you will.
We're geeks: let's all invest a little time to investigate the rhetoric. If you really don't like Flash, pick another reason: there are many ;)
And who gives a sh!t if Adobe makes contributions to open source? Contributions to open source don't make Flash open source.
ActionScript and Java are both published standards. Both have a canonical implementation which dominate distribution. Both are used as web plugins. Both standards are controlled by a single corporate entity.
Flash and Java are still free for end users and developers to use and distribute (alternative implementations are handled differently by Oracle & Adobe though).
The same can not be said for H.264. If i as a developer build an app that uses H.264 i have 0 confidence that i will not be sued in the future.
Flash's file formats are well-documented, but Adobe decides how it evolves. I can write my own Flash player today, but Adobe might decide tomorrow that they're going to completely change Flash 11's format so that my player is suddenly useless for any new content. They might have even decided it already, and after a year of development, the first I find out about it is on release day. Anyone who's invested time and effort into implementing my player in their software is suddenly screwed until I support the new format (which, because of things like proprietary streaming protocols, I might never be able to do).
This is actually the same problem with WebM. The source code is open, but it's not a community project in the way that something like Apache or PHP is. Google is the one in control, and this is Google's standard. If they're interested in openness, are they going to be submitting WebM to a standards body? Because unless there's a published standard that everyone (including Google) has to adhere to, WebM is even worse than H.264 because there's no guarantee that the money you've spent today will bring you anything tomorrow.
In other words, I have a hard time seeing how bundling a plugin and not supporting a proprietary codec that is vying to be part of a new standard is hypocritical at all.
And that above all seems to be the most important part of Gruber's "argument".
edit: minor grammatical correction
Plus I think it was adaptive quantization that gave the greatest single improvement.
Over both of those I think I'll take the implied legal opinion of the Google's legal team. In pushing WebM so strongly, Google obviously thinks it can withstand any attempted assault from the MPEG-LA. And unlike any of us or even Dark Shikari, they are lawyers who are qualified to make judgments about patent issues, and they know about exactly what hand Google is holding.
Second, virtually all of these questions take as an axiom that H.264 is IP-encumbered and WebM/VP8 isn't. Technical analyses (by people who have actually implemented H.264 and are familiar with the patents) suggests that that simply isn't the case
Possible, but if the issue came down to 'an h264 developer with the technical chops' vs 'The due dilligence exercised by Google when acquiring WebM for millions and millions', I think I would lean slightly towards the latter.
I mean that's been the recurring theme on that site re: corporate Google for years now. He's not throwing darts when Google does something in their own best interests. He's throwing darts when Google tries to spin those moves as being moves made in the interests of users, 'open development', free puppies and hugs for everyone, etc.
Come on. It's marketing. Too many black kettles and pots to count.
edit: second paragraph of parent post has been slightly edited since I posted this. Originally it only mentioned "open" without users, puppies, and hugs.
I feel the same way about Google. They spend a lot of their time trying to convince people they're the good guys, except that, like any corporation, they're wholly self-serving. I find it less than difficult to imagine them having reasons that make them feel like the good guys, while those reasons would make us feel like they're the bad guys.
In this case, they're trying to cram WebM down everyone's throats, supposedly because it's better for us. They're doing so by removing support for de-facto standards in their browser, and talking up a lot of rhetoric about free and open standards, but what they're really doing is trying to move from a format that a consortium controls to a format that they control. Until they submit WebM to a standards body, I can't believe that this is all for our benefit.
Everybody seems to be crying "but hardware support is coming soon!" without seeing it's difficult: "The unfortunate problem with this is that it’s a nightmare for hardware implementations, greatly increasing memory bandwidth requirements."
It comes down to WebM/VP8 being inferior (in specification and implementation) to an existing widely deployed (in bits and gates) codec.
Feel free to argue the fallacies of the osnews questions if your time is worth nothing.
Most people could probably agree that taking everything that John Gruber says as the truth without any other sources is a bad idea, but that's exactly the landscape of evidence presented for/against vp8. Thankfully, Dark Shikari's writeup is great, but there really needs to be more evidence.
Break it down: How many codec developers do you know? How many of those people do open source implementation comparisons? How many of those write it up? How many of those write it up intelligently for public consumption?
I wouldn't think that. You can't swing a stick without hitting a run-of-the-mill CRUD application developer — and judging from the articles that come across the HN feed it looks like everybody and their dog has invented a programming language — but video codec developers don't grow on trees. Besides, 6 months isn't that long at the pace that field seems to move.
Is that true for smartphones? I know that the iPhone 4, iPod Touch, and iPad (at least) support "H.264 video up to 720p, 30 frames per second, Main Profile level 3.1"
I don't know about Android devices.
It's one thing to be a proponent of the open web. However, it is just amazing the length people will go to in this "debate" to justify Google's hypocrisy. This is a corporate strategy move aimed at controlling the market and jamming up Apple and it's hundreds of millions of devices that play H.264. Gimme a break.
On the one hand you have Flash, which has been around for ages and you can't just take it out without breaking a lot of the web. If someone found an open way to render Flash (like if Flash Gordon somehow got good enough to do it), I guarantee you that Google would remove Flash. They would much prefer this to having to bend over backwards to render it in a different process and blah blah blah. The only reason Apple was able to remove Flash was because they entered the mobile landscape, where having a browser that couldn't render Flash still rendered more of the "real web" than most competing mobile browsers at the time. However, you don't see them dropping support for it in Safari (and not including the plugin on some macs is not dropping support). If you couldn't run flash at all on a Mac, ever, it would be disastrous.
H.264 on the other hand is a brand new technology, there is still time to make the right move before we end up in the same position as we are with Flash today: everyone wanting it gone but it not being easy to remove it (similar to the GIF situation as well). On top of that Google controls the biggest video supplier of the web, thus they have the additional ability to make the transition less painful.
Google does not need to bake Flash into Chrome. Somehow browsers have survived a decade without the need. Google's stance is bullshit corporate strategy. You bending over backwards to justify it is just comical.
> If you couldn't run flash at all on a Mac, ever, it would be disastrous.
For Adobe. It's not going to affect anyone's purchase of a Macbook just as it didn't affect anyone's purchase of an iPad.
> H.264 on the other hand is a brand new technology
Wrong. H.264 has been around for 7 years. It has hardware decoders in nearly every mobile device. It encodes 2/3 (and probably more now) of the web's video. It's used by almost every single major content provider.
It is WebM that is the new technology. Released 8 months ago to be precise. No hardware support yet (sorry, releasing VHDL designs doesn't count), almost zero content support, etc, etc. This isn't even close.
> there is still time to make the right move before we end up in the same position as we are with Flash today: everyone wanting it gone but it not being easy to remove it
Except no one wants H.264 gone. It's a high quality standard supported by devices everywhere.
> (similar to the GIF situation as well)
The GIF situation turned out fine. Thanks for the great counterexample to your logic.
Might I inquire where I can find Google Flash in Chrome? I only have Adobe Flash. Until you can show me a Google developed Flash, it's merely Google shipping an Adobe plugin, which is not the same as actively supporting H.264.
My personal fav:
If Apple were to switch to WebM and drop H.264 tomorrow,
would you then herald it as a great move?
The same couldn't be said if Apple dropped support for H.264. Old content would cease to work immediately.
That does not follow at all.
edit: I should say that it only follows if the patents in the patent pool apply exclusively/specifically to H.264.
The thing to remember is that when two systems each have patents, it's a false dichotomy to ask which set of patents trump the other. It's entirely possible that VP3 infringes on MPEG-LA patents _and_ H.264 infringes on On2/Google patents.
The chronology of the two is nearly irrelevant.
Where is the false dichotomy?
The current talking point on the h264 versus WebM war is that WebM is a patent landmine, while your back is covered with h264. There is some sort of broad belief that h264 is a universal got-your-back patent defense, and that is simply ludicrous.
As to your point, h264 has been on the market for a while now and is implemented by some seriously deep pockets. If you were the captain of the submarine patent ship, you would probably have fired the torpedos by now. If you're a new company, you'll feel pretty safe standing with Apple and all of the other MPEG licensees who have a vested interest in fighting. This is not the case with WebM.
And they don't. For example, "Method for processing decoded picture blocks in a block-based method of picture coding" was filed in '96 and granted in '99, and is part of the H264 patent pool.
There's a long list that goes like this.
MPEG-LA actually goes through great effort to show how each specific claim in the patent maps to a specific section in the spec: http://www.mpegla.com/main/programs/avc/Documents/avcCrossRe...
Mozilla is doing it entirely out of ideology (and sticks to their ideology much more than Google usually does).
[0] http://shaver.off.net/diary/2010/01/23/html5-video-and-codec...
Do you know how many engineers Mozilla can get for $5 million a year? Or how many ads they can buy for $5 million? Or how much beer? The point is- just because they have the money doesn't mean this is a good way to spend it.
Don't get me wrong there's already special deals baked into the fee format (like maximum fee caps for the big insider companies that could move the market on their own if you pissed them off) but it is in some ways "non-discriminatory", which is the ND in RAND, which would stop them doing a deal.
I bet it would help, but I don't see Apple doing that anytime soon.
Currently, Apple uses PowerVR and nVidia chips for hardware decoding in iDevices and Macs, and those companies have already shown interest in building chips that offer WebM support. But Apple would have to find some way to support WebM hardware decoding on current devices, something like a Rosetta layer for video playback.
If Apple were to only include full support on new devices, it would feel the wrath of the entire tech press, not to mention organizations like Consumer Reports and Greenpeace. Apple only risks that kind of bad press when it's about a technology they've authored themselves.
Moving from HTML5+H.264 to HTML5+WebM, or perhaps to supporting both, wouldn't be inconsistent with the switch from Flash to HTML5+H.264 they made with iOS.
On the latter, I'd estimate the risk that the MPEG LA would go after an MPEG 4 licensee (and member of the MPEG organization) for violating the patents they have already licensed to be fairly slim; one might expect them to focus more on non-MPEG-licensed vendors distributing WebM decoders and encoders. So if the power usage and component costs are reasonable, there should be little obstacle for Apple.
That would mean that (amongst other things) all video content from the iTunes Store would have to be reencoded for WebM and the YouTube app on iOS would stream WebM instead of h.264. But that way, older Apple products won't have access to the majority of online video, or if Apple provides software decoding, battery life will suffer significantly.
I can't imagine Apple even considering such a move, but if they did, don't you think the press would have a field day?
Not really, no. For example, h264 is hardware-accelerated on newer Macs (nVidia 320M / 330M), but not on the "old" nVidia 8600M-based Macs. The technical requirements are met (a 8600M is more powerful than a 320M), but Apple just won't release codecs.
A driver update isn't going to suddenly make iPads play WebM in hardware.
Similarly, Vorbis has been there for years. Why do you think Apple never supported it?
I don't. I think Apple will resist WebM for as long as they can. With h.264 support coming to Internet Explorer, I reckon Apple thinks h.264 is going to retain the majority of web users.
"Vorbis has been there for years. Why do you think Apple never supported it?"
Apple was the first to offer video on the desktop, I think they still consider themselves the authority. They have always preferred proprietary codecs (ALAC, AAC, h.264) over FOSS (FLAC, Vorbis, Theora), probably to keep control of their IP. I doubt that's going to change anytime soon.
The main benefit is that Google's move may reduce the risk that the MPEG consortium goes crazy in the future, and starts to charge every possible user, causing havoc in the process.
The main cost is more certain: a huge inconvenience for content producers that were hoping to get away with going H.264 only, causing havoc in the process.
The interesting question is the usual one: is the benefit worth the cost?
More - "why are you dropping support for H264 on grounds of openness but not dropping support for Flash (and MP3 and AAC) on those same grounds?"
All other browsers are open enough to get native support or have a plugin. Android phones will have support. Who knows or cares what happens on WP7...
Google is taking a stand. It is annoying that Google is removing existing support for h264. It will be annoying if Apple is completely against adding a software WebM decoder for current gen iOS devices, and ignore hardware solutions on next gen ones. But both companies are free to do whatever.
I also think this will have negligible impact in the real world. Youtube will always work on whatever. Content providers can just upload stuff there if they don't want to manage multiple formats themselves.
Google cant put hardware into most Android devices because they don't make them. Heck, you can barely upgrade your Android software on most carriers/devices. Google should first figure out how to prevent carries/manufactures from installing crapware on their handsets.
http://www.nvidia.com/object/tegra-2.html
Claims decode for Theora, VP8 and Vorbis, as well as encode and videoconference for VP8.
Basically every Android tablet at CES was Tegra 2 (I think it's the reference platform for Honeycomb) as well as a few of the high-end phones e.g. Motorola Atrix
http://daringfireball.net/linked/2011/01/12/slashdot-comment