WebM: Not Ready for Prime Time?
When Google made overtures to acquire On2 back in August 2009, the announcement was met with befuddlement by many, incredulity by a few, and praise from both the open-source community and On2 shareholders.
The latter group saw the purchase of On2, in part, as a vindication of a "revolutionary" technology. The Google acquisition, they reasoned, meant the sizable resources of an industry giant would be brought to bear against a technology that was locked within a smaller company but could go head-to-head against H.264, its primary codec competitor.
The most recent news in the On2-Google saga was the announcement by Google at its I/O conference earlier this month that it was "open-sourcing" code for both the last-generation VP7 codec as well as the newer VP8 codec, the latter of which Google and On2 both claim to be better than H.264, after VP7 fell far short in comparative testing.
With that context in mind, let's look at two of the three areas of Google's open-source video strategy: technology and business models. In a future article, we'll tackle the complexities of Google's licensing.
The biggest claim to fame for VP8 is as a replacement for H.264, which On2 made on September 13, 2008, under a press release titled "On2 VP8 Surpasses H.264, VC-1, Real Video in Quality and Performance."
Up until last week, the codec had never been properly compared against H.264 until last week. Prior to the independent testing by Jan Ozer and others, the claims of On2's management were the only basis that Google or anyone else had to compare VP8 against H.264. It turns out that the claims that VP8 is better than H.264 don't hold water, so a more in-depth testing of the claims is needed.
Nor does it appear the technology itself is ready to see the light of day. In conversations with several companies familiar with the situation, the code released to those who Google evangelized over the past few months since the acquisition was completed was described as fairly basic and even a bit counter-productive to true codec implementation. I was especially surprised to hear VP8 went through many code revisions in the last few months, with many base code revisions occurring in the few weeks prior to Google's announcement at the I/O conference.
What struck me as odd was not the fact that VP8 might need to have code revisions to refer to Google instead of On2, but that it needed significant modification this close to Google's open-sourcing announcement. In the 2008 press announcement, On2 claimed VP8 was ready for immediate usage by all major video player and compression tools.
"With On2 VP8, we set out to increase compression performance . . . while reducing playback complexity by 40%," said Paul Wilkins, Chief Technology Officer at On2, in the September 2008 press release. "In the end we exceeded these goals and even back-ported some of what we learned to our On2 VP6 or Adobe Flash encoders."
The one thing that Google did right in the release of WebM (the format that contains the VP8 video codec and the Vorbis audio codec) is to eliminate the older Ogg container, as well as Theora, the open-source video codec based on On2's VP3 codec.
Google states on its WebM blog that Ogg is now replaced by Matroska, which will house the video and audio streams of the .webm file extension. "A valid WebM file can only contain VP8 video and Vorbis audio in a .webm container. . . After much discussion with browser makers, tool developers and others, we reached a consensus that a narrowly defined format would cause the least confusion for users. If a user has a .webm file, they can be confident that it will play in any browser or media player that supports WebM."
Yet, for all the work being done to update the codec to make it somewhat competitive with H.264, including the flurry of code revisions over the past few weeks, the technology is still not documented like a true codec, regardless of whether it's open-source or proprietary.
Google seems more concerned with modifying its WebM FAQ than it is with helping the online video world understand the practical and financial benefits of an open-source competitor to H.264.