What happened to the Internet standards advocates who got everything they ever wanted? They lived happily ever afte— now, wait a minute. Microsoft, Apple – you guys better not play the Grinch on this one, ‘kay? Photo (CC-BY) love♡janine.

Shifts in standards usually take place at a glacial pace. This one may have just happened overnight.

Yesterday, the future of Web audiovisual content remained murky. You could choose H.264, a format that all but locked out open source software and threatened license fees down the road. Or you could opt for Ogg Theora, a format that was open but had serious patent questions around it, one major players said outright they wouldn’t support. Apple artificially emphasized choosing between Flash and the “open Web” – a misleading and false argument, especially since a common means of decoding H.264, by necessity, would become Flash.

Today, the landscape is different. If Google is a hundred-ton gorilla, it just took a belly flop of a dive into the debate at its Google I/O conference:


Initial, jet-lagged impressions, at least fed by my previous conversations with folks from the open video community:

VP8 video is really, truly open – free is in beer, freedom, standards, whatever. VP8 is a high-quality video codec, the newer generation of the codec incorporated in Flash. Yes, yes, H.264 can argue high quality – but it tends to be costly at both the decode and encode stage, at least with all the stops pulled. VP8 at least claims to be heavily optimized for this kind of delivery, and its new open licensing suggests the situation may only get better (see also: hardware partners below). That’s not “open” not with quotation marks around it, accompanied by some lengthy rationalization about the exact shades of meaning the word open can have penned by someone from a big company whose name begins with the letter “A.” (Either of them.) It’s open as in BSD-licensed, free, go build software that encodes or decodes it, go use it without fear of license fees today or a few years down the road. As for patent liability, anyone wanting to argue with this format will have to go toe-to-toe with half the industry, not to mention Google’s newly-expanded portfolio and deep-as-ever pockets and legal department.

WebM is open, top to bottom. Want audio, too? Try Ogg Vorbis, an excellent, efficient format free of the licensing requirements of MP3. Want a container format? That’s free and open, too, courtesy the WebM container, itself built on an excellent, open, extensible container already in use (Matroska MKV/MKA/MKS). Free source code? Check. Free encoding tools? Check.

…. with all the trimmings. Oh, you’re using FFMPEG (as do a number of VJ tools)? A patch for FFMPEG on any OS adds WebM support. Use Windows? DirectShow filters are ready. Parameters for encoding successful video in various use cases? Good God, man, these guys don’t mess around. The only thing missing right now according to Google is GStreamer support, but that’s coming soon, meaning I’ll soon even be using this format in Processing. You can even encode on the Web, with the commercial Zencoder product.

Updated – GStreamer support! WebM + GStreamer

And it’s got a posse. Sorenson Squish supports it. Google Chrome is onboard, but so are Firefox and Opera. YouTube has support via its HTML5 test, right now, today. Huge commercial video tool Brightcove, chipmakers ARM, NVIDIA, and TI, and many others (even Skype) are behind the format, too.

Adobe’s just fine, thanks for asking. Adobe is also listed as a partner. Expect full-blown Flash support, and robust encoding tools, too, I’d wager.

Who’s missing? Microsoft and Apple. Microsoft is already rumored to be supporting WebM and VP8 in IE9. Updated – Microsoft’s “all in” on HTML5, and supports VP8 – if you install the codec. The response is more than a bit tepid, but it doesn’t block people from using VP8.

Apple, your move. If you want an open Web, it’s just arrived. And since I hear Apple’s own H.264 patent stake is painfully small — and even bigger stakeholders spend more than they take in — I expect Apple to join in, too.

If it helps, Starbucks has unveiled new custom Frappuccinos, so maybe Eric and Steve can work out the final missing piece in the future of Internet video over a nice Venti or two. Go ahead, add whipped cream — the Internet has waited for this a long time. Treat yourself.

http://openvideoalliance.org/ has more informed analysis, as opposed to me; I just got off a transatlantic flight. And I expect they’re very happy indeed.

More coverage of how these formats work, and whether they’ll work for live visuals and not just video distribution online, soon — comments welcome.


Well, you know with something as sophisticated as video codecs, it can’t all be bad news. This analysis suggests some of the potential pitfalls:
[Diary Of An x264 Developer]

That’s just the sort of dose of reality I’d hope video developers and open video advocates wade through over the coming weeks, now that this announcement is formal and public. I expect such frank criticism may in turn be met with a defense of Google’s (and On2’s) choices, and development improvements in the long haul – and that’s healthy and necessary.

In short, the author, Jason Garrett-Glaser, “the current primary x264 developer and also an ffmpeg developer,” has some harsh criticisms for VP8 in terms of potential for patent liability, quality, and speed, though his thoughts are not without caveats and potential for improvement. The container format and audio codec, by contrast, are no-brainers.

It could be the jetlag, but my gut reaction remains: I think this is still big news, partly because of the number of partners Google appears in their corner – especially in terms of browsers. If either Microsoft or Apple joins, it seems like it’s a done deal, because it’s unlikely a lone browser developer would want to stand up against all the competition and YouTube. Apple in particular, having resisted even support for Ogg Vorbis on patent grounds (which, unlike Theora, seems to be a stretch), could rebel, though. In fact, if Apple makes the assessment that VP8 shares Theora’s performance, patent liability, and quality challenges, it’s hard to imagine Apple not protesting the move.

Patent liability still must be tested in the real world. I expect a full analysis from our friend Nilay Patel at Engadget who actually has an IP background. I don’t, so I’m going to just shut up now before I say something I regret.

But VP8 itself is not news. You can read On2’s previous claims about its awesomeness, whether you believe those claims or not. (Hint: encoding settings are everything.)

I certainly stand by the other aspects of my analysis, based on the previous contact I’ve had with some of the parties involved: don’t underestimate the will of these partners to find a solution, or the potential sway they can bring to the market here. Just having these browsers and YouTube and the code available today is already a major step.

If I’m wrong, I’ll make it up to you by posting a glitched-out video of me crumbling into little compression artifacts, crying about how I was made a fool. So, either way, you have a win-win situation.

  • Why do I get the feeling that the reason we didn't hear much of anything since the On2 acquisition was because Google really wanted to drop all of these bombs all at once?

  • This x264 developer thinks otherwise: http://x264dev.multimedia.cx/?p=377

  • Well, I actually have HTML5 video working in Firefox on my WindowsXP system now. As I type this, I'm watching a video on Youtube of a Gentle Giant performance. Even if VP8 will never be quite as good as h.264, the fact that it's completely open now more than makes up for that. But the fact of the matter is, it works right now. The only issue I can see off-hand is the current lack of full-screen mode.
    Okay, now I'm watching a Genesis video.

  • Okay, I did find one pretty big deal. The HTML5 stream of a video was soaking up about 40% of (one of) my CPU, whereas Flash video hovers around 8%. Clearly some work to be done yet. This is definitely still beta. But hey, it works, it's open now, and people can get to work on improving the codec.

  • Wow! I knew it, everything is solved with enough time! Well, kind of…

  • mgunes

    "The only thing missing right now is GStreamer support…"
    Not really.

  • Peter Kirn

    @mgunes: Thanks!
    @Darren: Well, the full-screen issue is a browser support matter; WebM is just the content, so it's up to the browser to use your OS' native fullscreen support and render that video on the GPU. And as for the CPU utilization, too many details there to know for sure, but for all the criticism of Flash, Flash is very mature. We'll just have to watch.

  • Oh, I have little doubt the technical side of things will straighten out in the next couple of months. I guess the real issue is the patent side. There are some pretty smart people working at Google, Adobe, and all those other places, though, and I can only hope they know what they're doing.

  • @Pete: WOW! You've got a Gravatar!!!

  • Y.

    While the x264 review is accurate (as far as I can tell), I don't find the criticism to be damning at all. I never expected VP8 (or any free codec) to be equal or superior to H. 264 Main Profile – given MPEG-LA's size and the state of the patent system this was always highly unlikely. VP8 only needs to be good enough to be competitve, given that large companies have to pay MPEG-LA for H. 264 use (In comparison, Ogg/Theora wasn't good enough and I'm glad the W3C didn't try to require it. Theora's shortcomings are known, but the container was way subpar in my opinion).
    Per x264's own comparison VP8 is as good as H. 264 baseline (using x264), which I presume is good enough for most people (that's why it's called baseline?), and I'm sure quality can be improved further. Most other criticisms are matters where it's reasonable to believe will be fixed/improved within a short time (e.g. spec ambiguity, CPU use of decoder), or patent fears (where x264 dev has no specific expertise at. I'm sure Google's lawyers know more than him).
    All that said, I am unsure we'll end up with VP8 being de-facto standard for <video> use. After all, H. 264 will keep its technical lead and mobile installed base. It should be sufficient though to convince MPEG-LA to adapt and keep reasonable license terms.