Can Google Reinvent Web Video?
The company’s free video format is set to be baked into phones and other gadgets.
An ambitious attempt by Google to shift the Web over to a new, royalty-free video format has taken significant strides. New software has been released that can build the format into dedicated chips for cell phones and other gadgets, perhaps the most crucial step before it can displace the proprietary video format that currently dominates.
Google’s video format is known as WebM. It was created by combining the preëxisting audio format Vorbis with VP8, a video format that Google bought last year with the intention of making it free for all to use in WebM. Google wants WebM to become the default for Web video and join the wave of new, powerful, and, crucially, free-to-use Web technologies such as HTML5 that enables Web pages to act like desktop applications.
But that would require displacing the well-established and proprietary H.264 format, now used for most streaming video online. H.264 is built into dedicated video chips in portable gadgets from phones to tablets and camcorders. A consortium called MPEG-LA controls the patents needed to create software or hardware that supports H.264. MPEG-LA levies a license fee on every unit shipped.
Enabling the development of equivalent chips for WebM is crucial if the rival format is to gain a foothold. Without such hardware, the work of encoding video is done by software that taxes a device’s main CPU too much, draining battery life. “The new hardware encoder encodes VP8, using a tiny fraction of the electricity that a general-purpose processor/CPU would use even at HD resolution,” says Aki Kuusela, engineering manager of the WebM Project hardware team. “This makes them very practical for mobile and other low-power devices,” he says. Without a dedicated video chip, such devices can typically muster only poor resolution.
The Google team tested the new hardware encoder by running it on simulated chips and real ones known as FPGAs that can be reconfigured to implement different hardware designs. Interested hardware firms can apply to receive the code for the new encoder online. Kuusela wrote in a blog post that “several top-tier semiconductor partners” are already starting to build their next chips with VP8 built in, but wouldn’t name specific firms. Major chipmakers, including AMD, Qualcomm, and Texas Instruments, are public supporters of the project, though, although they will likely support both VP8 and existing formats in their chips.
WebM has penetrated other parts of the Web’s ecosystem in recent weeks. The new version of the Firefox browser released last week has support for the format built in, while Google engineers built a software plug-in to add the same capability to Microsoft’s IE9 browser.
Yet almost a year since Google announced its WebM project, the format makes up a “very small percentage” of online video, says Jeff Malkin, president of Encoding.com, a company that converts online video between formats to allow publishers to serve different devices, and that has supported WebM since its launch, at Google’s request.
Being able to get the format built into hardware is a step forward, says Malkin, “but I don’t know if hardware is the bottleneck to it taking off,” he says, because WebM’s advantages remain unclear. It is much better than earlier royalty-free video formats, says Malkin, but so far the new format and the software Google has created to encode video in it can’t compete technologically with H.264. “It’s nowhere near the same on performance.”
Jason Garret-Glaser, who leads development of an open-source tool to encode H.264 video called x264, agrees. “The encoders for VP8 were initially terrible and are now less bad,” he says. A recent technical analysis that he was involved with found that Google’s encoders still can’t generate video as speedily or at the same quality as x264 can using the H.264 format.
With more work, it should be possible for VP8 to deliver performance close to that of the established, proprietary rival, says Garret-Glaser. But to make WebM stick, Google will need to leverage other assets.
This January, the firm removed built-in support for H.264 from its Chrome browser. But Google has been slower to make the most of its most potent lever, YouTube. Some videos uploaded to the site are converted to the WebM format. “But so far, YouTube appears halfhearted about it,” says Garret-Glaser. “Some videos encode, some don’t or take weeks.” He thinks that WebM’s destiny may be to be a competitor to H.264, not a conqueror of it.
However, “it’s nice to have that,” he says, “it keeps pressure on MPEG-LA, which may make its licensing more friendly,” he says. That should make it easier for developers to build on video technologies, creating new experiences for end users.
If such competition spurs video codecs to become more efficient, consumers will get other tangible benefits, says Eve Riskin at the University of Washington, whose research group has developed its own video encoders intended to enable video calling for sign language even on slow, non-3G cellular connections. “On a mobile device, the big challenges are battery life and bandwidth,” she says, “and better codecs can deliver more with the power and bandwidth you have.” The end of “all you can eat” mobile data plans by many networks could enhance the importance of good video compression for users, she points out.