How Will You Get Your Internet Video in the Future?

How Will You Get Your Internet Video in the Future?

Right now, the landscape for Internet video is relatively stable, with the vast bulk of online video sites (including the market leaders YouTube and Hulu) providing their streams via the Flash plug-in and usually containing H.264 encoded video.  Video shared on P2P networks is a different story of course, primarily MPEG-4 ASP in .avi or else H.264 in .mkv (although I am always surprised by the large amount of MPEG-2 content still traded today).  Even though the current situation is seemingly well established, a couple of recent developments have illuminated changes that are likely to have major impacts on the future of Internet video.  The aggressively anti-Flash moves of Apple in regard to iPhones and the coming iPad, coupled with the gathering momentum behind the html5 <video> tag, as demonstrated by recent experiments by YouTube’s and Vimeo, make very clear that Flash’s supremacy is not as ironclad as once thought.  At the same time, the html5 propagation, supported very ardently by open source and royalty-free software advocates, has exposed the troubles many see in the de-facto standardization of H.264 for web video, since that codec is contradictorily both an open standard *and* a patent-encumbered, licensed technology.  Some supporters of the html5 <video> feature over proprietary plug-ins such as Flash applaud YouTube’s experimentation, yet find the use of H.264 within html5 to be self-defeating at best and very troubling at worst.  They would prefer to see Google adopt a video codec that does not require a license, such as the open source Ogg Theora.  Many free software supporters therefore see the html5/theora combination as the perfect solution to the commercial and philosophical problems they see as unavoidable in an Internet video world dominated by Flash and H.264.

Much of the recent ardor on the part of these supporters comes from Google’s experimentation with html5, but even more so from Google’s very recently completed purchase of codec company On2.  On2 has been licensing its VP series codecs for years, including to YouTube, but its products have been overshadowed greatly by H.264, even as On2 marketers have been insisting that their soon-to-be-released VP8 codec will catch up and possibly even surpass H.264.  What does Google’s purchase of On2 mean for the dream of a html5/Theora based world, however?  On2 has already played a role in Theora’s history, as much of the open source codec is based on donated codec technology from On2’s VP3.  Naturally, some observers have speculated, therefore, that Google may take the codec assets of On2 it now possesses and donate them to the Theora project, giving it a much-needed boost of technology and development.  Such a step would come at a very opportune time for Theora, for it has fallen quite a bit behind the evolution seen by good H.264 encoders, such as x264 and  MainConcept, to the point were many experts in the field doubt Theora could ever truly catch up with H.264 from a quality perspective.

Sounds like a great scenario, no?  Google donates VP8, switches YouTube over to HTML5 exclusively using the newly improved Ogg Theora instead of H.264 and “voila!” … with a quick stroke Google eliminates the dominance of the proprietary Flash plug-in and the license-encumbered H.264, all at once.  Almost too good to be true, right?  Well, it probably is.  There are a large number of pending questions that greatly weaken the likelihood or feasibility of such a Google master-stroke.  First is the question of just how eagerly video sites, including YouTube, will embrace HTML5 as there is, as yet, no DRM element built into it as there is with Flash.  While that may sound great to viewers, it could cause any site with hopes of selling or renting premium video content to avoid HTML5, including YouTube and almost assuredly Hulu.  Second, and far more importantly is the issue of whether Ogg Theora, even with an infusion of technology from the On2 purchase can ever really present comparable quality to H.264.  Right now even the Theora developers at the Xiph foundation are very circumspect about making ultimate quality promises.  Other voices in the video codec community are far harsher, blasting both Theora and the Ogg container as just not serious competitors currently, and not likely to ever be.  Third, the claims that On2 has made about VP8’s patent insulation are seen very skeptically among those in the know.  I’m far from a patent lawyer (thankfully!) but there is almost no field of inquiry that is more of a patent minefield than video compression today.  In fact, there are some real concerns about the patent claims of the MPEG-LA over H.264!  In other words, almost anything Google does with VP8, especially something as momentous as “taking it open source” is likely to just create a more treacherous legal landscape that what exists even today.

So, where does that actually leave us today?

H.264: highest quality/best implementations, most widely compatible, widespread hardware adoption, free for decode through 2015, probably patent-proofed, but with a large cloud of uncertainty over its licensing future after 2015.
Ogg Theora: moderate quality, almost no software implementations, no hardware adoption, ostensibly patent and license-free, but with large cloud of litigation uncertainty hanging over it.

My analysis is that Google has no short term intentions of open-sourcing VP8, but will use the technology they acquired from On2 as leverage when negotiating with MPEG-LA over the licensing costs of H.264 in the future.  Ogg Theora will benefit from the enormous amount of enthusiasm and passion open source advocates are bringing to it, but that ultimately as a competitor to H.264, Theora will not get very far.  Codec development is very, very difficult stuff, and there just are not that many people around that can do it well.  HTML5 will grow in acceptance, but will exist alongside Flash for a long time to come, at least until Hollywood wises up and ditches DRM.  Of course much of this would be a moot point if we all got smart and did away with software patents, but again, that is more wishful thinking than a likely outcome.