Google I/O 2010: WebM - A Technical Overview and Possible Legal Issues

In the first day of the Google I/O 2010 developer conference, Google unveiled WebM, a new open and free video format aimed at solving the HTML5 video, codec debate. We’ve already covered the details of the WebM Project, the coalition of companies that support the new format, which includes Mozilla, Opera, Adobe and many other relevant companies.

Everything looks good on paper and, at face value, WebM is set to take over the web. With support from most web-browser manufacturers, software vendors and chip makers, what could go wrong? Well, let’s take a look at the more technical details of the WebM video format and later at the legal issues that could arise.

The three components of the WebM video format

Technically, WebM is made up of three components. One is the VP8 video codec developed by On2. The On2 video-compression technology company was acquired by Google earlier this year. It has a long tradition in the market and its products are generally well regarded. VP8, the company’s latest product, is a high-quality video codec, more on that aspect later, and Google is releasing it with a BSD-type, royalty-free license. This means that anyone will be able to use, modify and distribute it freely.

The second component of the WebM format is the audio codec. With this, Google went with the tried-and-tested, open-source Vorbis audio codec, which has been around for years. It was the only real, viable choice for WebM. It is broadly supported on most platforms and its performance and quality are well known.

Finally, the team behind WebM, chose a container format based on the existing Matroska, an open-source multimedia container. Again, it’s a well-known technology and is suitable for the type of tasks WebM is intended for, especially streaming. Matroska supports ‘real’ streaming, as opposed to most other popular choices online today.

Does VP8 have what it takes?

The only real unknown here is VP8. On2 has stated that it stacks up well against H.264 and everyone is quick to tout the codec’s performance and quality. However, an early review of the VP8 specifications, which Google says are final, reveals some problems.

Jason Garrett-Glaser, developer of the popular x264 open-source H.264 encoder, managed to get a hold of the VP8 specifications, source code and official software ahead of the announcement. He posted a detailed overview, which makes for an intereresting read, covering all of the technical aspects of the newly open-sourced codec and things don’t look that great and certainly not as clear cut as others would have you believe.

VP8 quality and performance in comparison to H.264

From his review of the specifications, the developer concludes that VP8 is a very capable video codec but not on par with H.264. It is better than Theora and all other open choices out there and comparable with the H.264 Baseline Profile in terms of quality. However, in his opinion, it is not comparable with higher quality H.264 profiles developed for HD videos and so on.

He believes some of the issues could be fixed, but since the specifications released by Google are supposed to be final, it creates an issue. Hardware manufacturers need a stable implementation, so things are unlikely to change anytime soon.

In terms of decoding performance, VP8 seems to be on par with H.264 and the fact that it will get support from hardware makers means that this will translate into real-world performance as well.

So it would seem that VP8 is not as good as H.264, overall, but it’s good enough to be usable, when weighing in the fact that it’s open and free. However, there may be problems here too. Again, from Garrett-Glaser’s overview, it would seem that VP8 and H.264 are very similar in a number of ways. So similar in fact that they open the door wide for any number of patent lawsuits.

VP8 may be under threat of patent lawsuits

Google may be releasing it under a liberal license, but it seems even the tech giant can’t escape the problems with software patents today. This case underlines the huge problems software patents lead to. A codec is more or less a glorified algorithm, or rather, a number of algorithms. In general, some algorithms are going to be better than others at solving one particular problem. This is why there are no patents on mathematical formulas, some problems have just one viable solution. Yet, software companies are forced to devise ways of solving the same problems differently, just so that they don’t get sued.

It will be interesting to see how this plays out for Google. It’s certainly a powerful company and it has much of the industry on its side. Microsoft is open to the WebM idea, so there should be no problems there. Apple, on the other hand, has stayed quiet. Apple has so far supported just H.264 and has shown no sign that it will support VP8 or WebM. The company, along with Microsoft and many others, is a licensee of H.264 and part of the MPEG LA, which oversees the codec.

An attack from MPEG LA, or other companies or groups that may hold patents relevant to video compression technology, would likely mean a legal battle that would drag on for years and cost everyone involved vast amounts of money. Google is deep-pocketed enough to handle it, but other smaller companies are not.

All in all, WebM holds a lot of promise. It’s technically good enough to stand up to H.264, it has been released under an open license and enjoys the support of a lot of key players. But it’s too early to tell if this will be enough for it to be a success. The coming months and reactions from other parties with interest in the market will make or break Google’s bold move.