Monday, February 14, 2011

Royalty-free MPEG video coding standard ups the ante for Google's WebM/VP8

Amid all the debate about whether Google's WebM video codec VP8 is truly patent-unencumbered and a valid alternative to the market-leading video codec (AVC/H.264), a very interesting project nearly went unnoticed: the development of a royalty-free MPEG video coding standard.

At a recent meeting of the Motion Picture Experts Group (MPEG), the decision was taken to issue a call for proposals (which will formally happen in March 2011) on video compression technology "that is expected to lead to a standard falling under ISO/IEC 'Type-1' licensing, i.e. intended to be 'royalty free'." The announcement also states the following:

"In recognition of the growing importance that the Internet plays in the generation and consumption of video content, MPEG intends to develop a new video compression standard in line with the expected usage models of the Internet. The new standard is intended to achieve substantially better compression performance than that offered by MPEG-2 and possibly comparable to that offered by the AVC Baseline Profile."

It's important to note the difference between MPEG, the standardization body, and MPEG LA, the licensing agency. MPEG defines standards, and those can be royalty-free or royalty-bearing, depending on what the owners of the essential patents agree upon in each case. MPEG LA manages patent pools related to royalty-bearing standards, serving right holders and licensees as a one-stop solution by way of aggregation.

Rob Glidden, a former Sun executive and digital TV expert, made me aware of MPEG's royalty-free codec plans via Twitter (thanks for that!). I found several interesting posts on his blog, such as this recent one arguing for a royalty-free video codec standard. Just like me, Rob appears to be somewhat skeptical of claims that WebM/VP8 is truly patent-free.

Three formidable hurdles for WebM/VP8

Google and its WebM allies (primarily Mozilla, Opera and the Free Software Foundation) wanted to pose a direct challenge to AVC/H.264, the incumbent and ubiquitous video codec. Their strategy was to seek adoption of WebM/VP8 as an Internet standard, claiming that it was the only format of its kind to be available on royalty-free, open-source-like terms. They thought they were already challenging the champion in the final match, but not so fast: MPEG LA may soon start collect royalties from VP8 adopters, and if MPEG's own royalty-free codec initiative succeeds, VP8 will have to beat it -- or be forgotten.

Using that tournament analogy again, VP8 is only in the quarter finals. MPEG LA's formation of a VP8 patent pool and MPEG's call for proposals on a royalty-free video codec are such important developments that the results they produce will have to be awaited before standardization on VP8 will even be seriously considered by the W3C or IETF. There is no rush because everyone can demonstrably watch Internet videos for the time being even though HTML 5's video tag is currently codec-agnostic.

I have drawn up a simple graphic that shows the three hurdles VP8 is facing:

If MPEG LA's patent pool for VP8 disproves claims that Google's codec is truly free of charge, it loses its intended unique characteristic and is out of the game. At that stage, even the Free Software Foundation may feel forced to withdraw its support. It wouldn't be possible for the FSF to advocate a royalty-bearing format.

Should VP8 overcome that hurdle, it would play a semifinal match against MPEG's royalty-free standard. I'll explain further below why this one is also going to be a tough challenge.

Only if VP8 survives that round as well, it will get to challenge AVC/H.264. To dethrone the reasonably-priced market leader, VP8 would have to deliver high quality, not just royalty-free availability.

The odds are long against WebM/VP8

Surmounting three significant hurdles in a row -- since failure at any of them means "game over" -- is really difficult. Even if one assumed a 60% chance in each case, the aggregate likelihood of success would just be slightly above 20%. In my opinion, VP8's chances are less than that. Let me explain my view on each of the three challenges.

1. MPEG LA patent pool for VP8

MPEG LA requests submissions of patents essential to VP8 until March 18, 2011. That's less than six weeks from the announcement. I presume MPEG LA already knows of a variety of relevant patents, and has discussed this with its membership. The first time MPEG LA CEO Larry Horn mentioned the possibility of such a pool was in an interview with AllThingsD in May 2010.

After the submissions deadline, MPEG LA's patent evaluators will make a determination as to which patents are truly essential to VP8. Obviously, whatever they conclude isn't the same as a judicial declaration of infringement. But MPEG LA is trusted by industry, and the moment it specifies the patents it believes read on VP8, Google and its allies will either have to invalidate those patents (which is unlikely to work for a reasonably long list) or prove that there isn't an infringement.

There could also be patents which VP8 infringes but which aren't submitted to MPEG LA in that process. I didn't mention that before, just to keep things simple at the time. When assessing the likelihood of VP8 truly being patent-unencumbered, this additional risk must, however, be taken into account.

2. Royalty-free MPEG video coding standard

It's plausible that MPEG can pull off a royalty-free coding standard. It will be able to draw from three types of patents for that purpose:

  • patents that have already expired

  • patents approaching expiration (which limits their commercial value)

  • patents whose owners are willing to forgo short-term revenue opportunities in favor of longer-term objectives, or who consider it a prudent strategy to support a royalty-free standard for web purposes and monetize related (but other) patents in other fields of use such as digital cameras and digital television

Codec development was already reasonably advanced in the early to mid 1990s, and US patents of that kind have a maximum term of validity of 20 years. Therefore, the first two groups of patents may already be sufficient to produce a codec at a level with, or even significantly superior over, VP8.

In order to compete successfully with a royalty-free MPEG standard, VP8 would have to be clearly better just to be considered. That's because MPEG has the backing of many vendors and offers an open and inclusive process. Under MPEG's stewardship, such a standard can continue to be developed (as new patents expire, approach expiration or are made available by their owners for other reasons) in a balanced, consensus-based way, while WebM is dependent upon Google, a company with specific strategic interests.

The argument that the availability of a WebM reference implementation on open source terms ensures vendor independence isn't sufficent. Theoretically, it's true that anyone could take it in a different direction. Practically, such forking would result in fragmentation, and it's doubtful that anyone would ever put significant resources behind such an effort.

3. Incumbent AVC/H.264 standard

Should VP8 really overcome the challenges previously described, it would still have to garner broadbased industry support. It wouldn't be in the interest of standardization bodies like the W3C and the IETF to formally declare a "standard" that fails to be adopted widely. Such failures can undermine the credibility and diminish the influence of those organizations.

AVC/H.264 isn't royalty-free, but its terms appear to be acceptable to a majority of industry players. In order to take any sizable part this incumbent's market share away, VP8 would have to be at least technically comparable. Degradations wouldn't be acceptable to consumers. However, my own impression and that of most of the opinions I read is that there's still a noticeable gap.

One of the challenges for VP8 is the limited number of hardware products supporting it. For hardware companies it's particularly important to be careful about patent-related risks. A software company may be able to just change its code. Hardware companies would be hit hard by injunctions or ITC import bans. VP8 faces a chicken-and-egg problem with the hardware industry, and the two new initiatives I described before will also have an effect in that area.

Google and its WebM allies won't give up. They are determined to turn this into a new edition of the "videotape format war" of the late 1970s and early 1980s. Looking at the drawing that shows the three hurdles and considering the facts, I have my doubts that they will get very far. However, even if the odds appear to be long against VP8, it may have a positive effect on the willingness of patent holders to support MPEG's royalty-free codec standard.

If you'd like to be updated on the smartphone patent disputes and other intellectual property matters I cover, please subscribe to my RSS feed (in the right-hand column) and/or follow me on Twitter @FOSSpatents.

Share with other professionals via LinkedIn: