For most of its life (since the late 1990s), the H.264 video coder specification (aka MPEG 4, class 10) has been the video codec compression/decompression algorithm of choice in most consumer embedded and mobile devices. At last count, about 50 companies supply the H.264 hardware/software building blocks necessary for building video-capable devices and systems.
A few years ago, that domination appeared to be threatened by Google's introduction of its open-source WebM video coder , which Google was recommending to hardware and software developers building Android-based mobile and consumer video platforms. That domination seemed all the more tenuous with the availability of several other open-source video coder alternatives, among them Dirac, FFMpegand Theora.
But at the 2013 Consumer Electronics Show last month, while I might have
missed them in the 100s of exhibits and demos, I could not find any video-enabled mobile, desktop or embedded consumer device or system using those open-source alternatives.
"I could not find any video-enabled mobile, desktop or embedded consumer device using those open-source alternatives."
What counts as such a device?
Web developers are all moving to HTML5. One major attraction is the VIDEO keyword, which allows embedding video without the use of Adobe Flash.
But you still need a codec to show video. YouTube had a beta site that used HTML 5 instead of Flash, and it worked in IE and Chrome, but not Firefox. The issue was licensing: the YouTube site used H.265. IBM and Google had paid the license fee to use the video codec in IE and Chrome. Mozilla did not. The issue wasn't the fee, but the commitment to open source. Mozilla needed *everything* to be available in source form if desired, and the H.265 codec could not be. The license did not permit source distribution.
Google subsequently announced a commitment to make Chrome open source, and that it would end support for H.265 in Chrome. It felt open source alternatives like WebM had achieved levels of performance that made them competitive.
I expect H.265 to have a market in the embedded and mobile device spaces, but not in places like browsers where the availability of source for everything shipped with the browser is a requirement.
I think the main issue in these codec wars is only about licensing. From what I've seen in the past, the proprietary codecs (open source or not) use the same algorithms generically as the ITU blessed ones, with just enough variation, or feature purge, to allow them to sidestep the licensing issue. So I don't expect any (factual) technical superiority from the alternatives.
What makes H.265 interesting to me is that compared with MPEG-2 compression, i.e. H.262, it offers enough of an improvement as to make UHDTV feasible for the masses. As opposed to UHDTV being a cool oddity you see only at trade shows. And what makes UHDTV interesting is that at long last, here comes a TV standard that can actually match the quality of excellent camera lenses. The lens now being the next weakest link in the chain. Until now, even HDTV did not come close to exploiting what image info a good lens passed through.
The licensing arrangements will make or break H.265.
David Patterson, known for his pioneering research that led to RAID, clusters and more, is part of a team at UC Berkeley that recently made its RISC-V processor architecture an open source hardware offering. We talk with Patterson and one of his colleagues behind the effort about the opportunities they see, what new kinds of designs they hope to enable and what it means for today’s commercial processor giants such as Intel, ARM and Imagination Technologies.