May 19, 2010
By Jan Ozer Contributing Editor
Online Video News

Google Open Sources VP8

[Updated Thursday, May 20, 8:15 a.m.ET]

Today, at their Google I/O Conference, Google announced that they would open source the VP8 codec that they acquired along with On2 in a deal that closed early this year. By the time the announcement was made, it had been predicted for so long and so widely that it was as anti-climactic. I approached the announcement with a list of questions. After covering the basics, we'll see which have been answered, and which haven't.

By way of background, Google announced that they were fully open sourcing the VP8 codec, so that it will be totally royalty free. They also announced a new container format for VP8 video called WebM, which is based on the Matroska media container, which can only contain VP8 video and Vorbis audio. All Webm files will have a .webm extension (appropriately enough), and will play in HTML5-compatible browsers updated to support WebM; see http://www.webmproject.org/users/ for the list.

There won't be a test, but to avoid confusion, VP8 is the video codec, WebM the container format.

Those are the basics, the rest you can read about at www.WebMproject.org. Let's get to my questions.

Who's the supporting cast?
No technology—even from Google—is an island, and even Google needs broad industry support to succeed. So the first question I had was who's supporting it.

Getting vendors to support technology announcements like this is about as hard as giving free beer away at frat party (at least the ones I went to). The depth and duration of support will depend upon how VP8 succeeds in the marketplace long term, which is probably likely, but certainly not inevitable. That said, who signed up for the VP8 launch party?

Beyond Mozilla and Opera, lots of folks. Let me give a quick shout out to Sorenson Media, who supplied the comparison VP8/H.264 test clips I'll discuss in a separate article (so I know their Squish tool is working now); on the encoding side, they were joined by the usual suspects, including Telestream, Rhozet, ooVoo, and many others. As a nice surprise, Adobe CTO Kevin Lynch spoke at the event, primarily relating to HTML5 support in Dreamweaver and other development tools, but also regarding Adobe support in VP8 for another little Adobe application discussed below.

Hardware vendors are a little bit tougher to acquire than software, but tend to sign on even if they have nothing specific to announce. For example, NVIDIA's name is on the list of supporting hardware, but here's their illuminating quote. "As a leader in visual computing, NVIDIA is always looking to enable users with the best video experiences, whether it's creating, viewing or broadcasting. We support Google's efforts to drive broad adoption of the VP8 codec and new WebM format." Falls something short of "we will support VP8 playback on all future NVIDIA graphics chips, eh? AMD (who owns graphics card vendor ATI, NVIDIA's closest competitor) was also on the list, but didn't have a separate release.

ViewCast announced that "Starting the third quarter of this year, Niagara SCX will enable users to capture and stream video utilizing VP8—the high performance, open source video codec that contributes to the WebM project under a royalty free license." Inlet also pledged support, but didn't provide a due date "we are building VP8 support into our entire suite of solutions, including our Spinnaker live streaming appliances and Armada high-volume transcoding system."

In the mobile arena, Qualcomm's release stated "We thus continue to collaborate with On2/Google's engineering teams to support VP8 codec on our mobile platforms and deliver a rich video experience on Qualcomm-powered mobile devices. Texas Instruments chimed in "With access to the VP8 code, our OMAP 4 platform delivers high-resolution VP8 decode at the low power levels that mobile architectures demand. We deliver VP8 performance efficiently, leveraging a highly differentiated video engine as opposed to complete reliance on the MCU." So mobile phone support from at least two vendors in the space appears to be in the works.

On the platform side, Kaltura, Encoding.com, Brightcove and others also announced support for Webm, with many more likely on the way; the full list is at http://webmproject.blogspot.com/

At this point, we know that real-time encoding hardware will be available, presumably sometime soon, but don't know if GPU-accelerated playback is in the cards, though it's likely a question of if rather than when. This is a significant issue, as one of the questions in Google's FAQ was:

WebM playback seems to use a lot of processor resources on my computer. Why is this?

The Developer Preview releases of browsers supporting WebM are not yet fully optimized and therefore have a higher computational footprint for screen rendering than we expect for the general releases. The computational efficiencies of the VP8 codec are more accurately measured today using codec-level development tools in the SDKs. Optimizations of the browser implementations are forthcoming.

Optimizations are nice, but as we saw with Flash and H.264 in Safari on the Mac, GPU acceleration is the way to go.

What about Microsoft and Apple?
Both conspicuously absent. In Apple's case, this means we don't know if iDevices or the Safari browser will support VP8. Ditto for Microsoft devices and Internet Explorer, which currently sit at between 40-50% market share.

Can they not adopt VP8? Interesting question. As I understand it, both sit on the WC3 committee that decides what comprises the HTML5 standard, and both could say no, just like they did with Ogg Theora. On the other hand (see next question), Google can provide significant leverage via YouTube to force both Apple and Microsoft into the fold.

How much will Google use YouTube to "incent" Microsoft and Apple to support WebM?
This is the stick, eh? A browser that doesn't play YouTube is like (to continue a theme) a frat party without beer, which is to say not very popular. If YouTube went VP8 only, both Microsoft and Apple would have to support the new technology; Google fell short of that position, simply stating something like that they were going to reencode their entire HD catalog into VP8 format. Makes you wonder how many servers YouTube actually has, but that's what they said.

Will Adobe include VP8 playback in Flash?
Yes it will, Adobe's Kevin Lynch announced to the polite applause of the HTML5 centric crowd. Does this mean Flash isn't, as so many have been predicting, dead?. Not sure; give me a couple of weeks to sort through this one.

What about Android?
Android will certainly support WebM via Chromium, Opera, and Firefox. But Lynch also said that Adobe had some Android-related announcements for day two of the event. Remember that Flash Player 10.1 will have hardware acceleration on the Android platform, at least for H.264.

What about MPEG-LA (which represents the H.264 patent group)? Could this cause the patent group to extend the non-royalty policy for free internet use in perpetuity?
I sent a late note to MPEG-LA, and haven't heard back, but that's more my fault for being late than theirs. It will be interesting to see when and if MPEG-LA responds to this announcement.

And finally, the big one? How does VP8 quality compare to H.264?
Check here to see for yourself.

UPDATE:

In an announcement several hours after Google's, Microsoft indicated that IE 9 will support VP8, but only if its already installed on the system, so it won't include the VP8 codec with the IE9 browser like Google, Mozilla, and Opera. This is an odd choice, since the technology is free. Though there could be any number of reasons that Microsoft would make this decision, one prominent one is that they feel that VP8 might infringe on technology patents owned by the H.264 intellectual property group administered by MPEG-LA.

As you may know, Microsoft is a member of the MPEG-LA patent group, as is Apple, and concerns about submarine patents was a frequently stated reason that Apple declined to support Ogg Theora in its bid to become the designated HTML5 codec. Hmm, could history be repeating itself?

Interestingly, we asked MPEG_LA for a response to the Google announcement; you can read their comments below. Suffice it to say that they didn't give VP8 a clean, patent-related bill of health.

"There are likely many patents used, and users have an interest in knowing the cost and availability of licenses under them," says MPEG-LA's Tom O'Reily. "Therefore, a patent pool license would be of service to the market so users may build the cost of using the technology into their business models and gain appropriate permissions to use the patents on reasonable terms if they decide to do so."

For clarification, we asked ‘are you talking about VP8 or H.264? Are you raising the specter of patent claims on VP8?" MPEG-LA responded "Thanks for checking. We are talking about VP8. We do not have any additional comment beyond our initial response."