Cisco's Thor project swings a hammer at Web video codecs

Cisco is the latest big name to take on the challenge of developing an unencumbered video compression technology

With its Thor project, Cisco is trying to provide the tech industry with a royalty-free video codec for use in Web browsers and online video, an issue that has stymied the industry for years.

Video codecs provide compression algorithms to reduce video size so that they can be more easily transferred via the Internet, with minimal loss in image quality. But these codecs, such as the widely used H.264 and its successor H.265 (both known as HEVC), are covered by patents, so browser makers and other tech vendors need a license that may include royalty payments and/or restrictions on the codec's use or distribution.

When the HTML5 specification was up for standards approval, the W3C (World Wide Web Consortium) sought out a video codec but ended up not placing one in the specification because of license issues. Google tried to provide an alternative to the patented video codecs with its VP8 codec and, later, the VP9 version, as did Mozilla with its ORBX.js codec. However, neither were adopted due to other browser makers' resistance to supporting a competitor's technology or concerns over licensing terms.

New patent trolling spurred Cisco to act

Cisco is trying its hand at solving the video codec conundrum under the auspices of the IETF (Internet Engineering Task Force), rather than the W3C. But Cisco CTO Jonathan Rosenberg admits the effort is in a "super-early" stage of development and won't be ready for several years: "It is our objective to be as good if not better than H.265."

Why bother? Because H.265, the successor to the H.264 codec around which Cisco standardized its WebRTC products, has become the target of patent trolls seeking to get more royalties. "Unfortunately, the patent licensing situation for H.265 has recently taken a turn for the worse. Two distinct patent licensing pools have formed so far, and many license holders are not represented in either.... The total costs to license H.265 from these two pools is up to 16 times more expensive than H.264. H.264 had an upper bound on yearly licensing costs, whereas H.265 has no such upper limit," Rosenberg has written. The licensing terms also preclude use of H.265 in freely distributed software such as browsers, he says.

Currently, Cisco is paying $6.5 million per year for use of H.264, Rosenberg tells InfoWorld, so the huge leap in H.265 royalties and its lack of an annual cap, is very worrisome. And Cisco won't use Google's open source, royalty-free VP8 codec because it was developed solely by Google, he says, making it a risky technology on which to build products.

The Thor project has its first draft release -- and backers

The Thor initial draft was released to the community about two weeks ago, via open source, Rosenberg notes. Cisco has submitted it to the IETF, which had already begun a standards effort in its NetVC workgroup to develop a next-gen, royalty-free video codec based on Mozilla's Daala technology. Cisco figured it made sense to have a single effort to solve the video-codec licensing problem.

"There's an industry need for codecs of this type, and Thor and VP9 seem to be [credible] alternatives to HEVC," says Iain Richardson, the principal at codec consultancy Vcodex. "But it's early."

IDC analyst Al Hilwa says Thor is a "great move by Cisco." But he wishes Cico and Google would combine their Thor and BP9 work: "It would be great to see some confluence in these efforts."

Mozilla is backing Cisco's endeavor. Mozilla already has its Opus audio codec effort under there auspices of the IETF. "We think Mozilla's Daala and Cisco's Thor form a strong basis to repeat that success in video." Cisco's Rosenberg echoes those sentiments, saying Cisco would be happy to leverage the best of Thor and Daala.

Join the Computerworld newsletter!

Error: Please check your email address.

More about CiscoGoogleIETFInternet Engineering Task ForceMozillaW3CWorld Wide Web Consortium

Show Comments