The State of Video Codecs 2016
2015 was a tumultuous year for codecs. HEVC seemed like a sure thing in January, looked to be in trouble in July, but was happily resuscitated in December, albeit with some missing productive months from its useful life. Open source codecs began the year with the entropy normal only in the codec world, as entities with clearly aligned goals evolved in completely different directions. Then, in the rarest of occurrences, the companies reversed course and converged toward a single standard, which almost certainly helped shove HEVC back on its intended path. Along the way, there were some interesting new technologies making the usual bold (and unproven) claims of pristine quality at seemingly impossible data rates (PERSEUS, Tveon). In case you missed it all, here is your codec year in review.
At the start of the year, HEVC turned 2 years old, and there was a single royalty pool from MPEG LA that charged $0.20 per unit for encoder and decoder after the first 100,000 units, with a $25 million cap in the first year, and no royalties on HEVC-encoded content. Several services had already deployed HEVC content, most famously Netflix with House of Cards.
On the open source front, Google was delivering about 30 percent of all content via VP9 to save big on bandwidth and improve the quality of lower-bitrate files. In October 2014, the Internet Engineering Task Force (IETF) sponsored the Internet Video Codec (netvc) working group, with a goal of delivering the specification for a competitive open-source codec by May 2017. One proposed codec was Daala, a collaboration between Mozilla and Xiph.org. Daala was making progress, but it wasn’t close to commercial availability. The other candidate was VP10, the next-generation codec after Google’s VP9.
On March 26, HEVC Advance, an independent licensing administrator, announced a new HEVC patent pool with an initial list of licensors that was expected to include GE, Technicolor, Dolby, Philips, and Mitsubishi Electric. The new company promised to announce royalty rates and licensing terms in the second quarter of 2015.
Happy people on the HEVC Advance website. Clearly they weren’t discussing feedback from the company’s first proposed licensing structure.
On April 1, V-Nova, a London-based company founded in 2011, announced the new PERSEUS codec, which, according to the press release, offers “UHD quality at HD bitrates, HD at SD bitrates, and SD video at audio bitrates.” Streaming Media has asked repeatedly to test PERSEUS, and V-Nova has declined, though before IBC, the company claimed that it was “working with a number of operators” to trial and evaluate Ultra HD using its PERSEUS codec.
On Nov. 23, 2015, V-Nova announced that Spanish broadcaster MediaPro Overon deployed a PERSEUS-powered product to deliver a contribution stream from a soccer stadium to the Overon control room. According to the release, PERSEUS delivered an Ultra HD stream at 240Mbps with “visually lossless” quality.
Certainly, proprietary technologies such as PERSEUS have the smoothest path toward success in closed-loop markets, and it’s good to see V-Nova making progress. But neither the closed-loop application nor data rates presage PERSEUS’ release—at least not anytime soon—into general purpose streaming markets, where data rates are much lower, and details—for instance, licensing and compatible decoders—complicate productization and market penetration. At this point, PERSEUS remains an outlier—neither standards-based nor open source—and without third-party tests, it’s difficult to determine its potential.
V-Nova was launched on April Fools’ Day. This, and the company’s aggressive claims, gave many compression pundits pause.
Breaking the Camel’s Back
On July 21, HEVC Advance announced its licensing terms. Briefly, encoder/decoder pricing, whether in hardware or software, ranged from $0.80 per unit to $1.50 per unit, with no de minimis exceptions and no caps, backdated to the technology’s first use—take that, you pesky early adopters! HEVC also claimed a 0.5 percent royalty on attributable revenue, essentially a royalty on HEVC-encoded content.
Industry reaction to the proposed royalty was almost uniformly negative on multiple fronts.
Writing for the Streaming Media blog, Dan Rayburn pointed out that HEVC Advance is long on claims but short on details. Potential licensees also vehemently rejected the content royalty. For example, Joe Inzerillo, the CTO of MLB Advanced Media, told Fierce Cable, “The notion of gross revenue is the part that’s just a non-starter. If it was just another toll to pay and the toll was sort of reasonable ... that wouldn’t be such a big deal. But the gross revenues ... no mainstream company is ever going to do that.” Inzerillo went on to say, “The H.265 patent stuff has really taken the wind out of 4K sales and 8K.”
On the encoder/decoder side, it was also unclear whether HEVC Advance’s proposed royalty rates, which exceed those proposed by MPEG LA by as much as 10 times over, would withstand a FRAND (Fair, Reasonable, and Nondiscriminatory) challenge as required for technologies used in an industry standard. With companies such as Apple looking at a $120 million first-year payout, a FRAND challenge seemed almost assured.
Finally, as Rayburn wrote in the blog post mentioned above, a third HEVC licensing group was also forming. Suddenly, HEVC morphed from fait accompli into an unmanageable mess. Meanwhile, behind the scenes, the open source community was making plans to consolidate its efforts. But first, there was Thor.
On the Open Source Front
On Aug. 11, Cisco announced Thor, an open-source codec the company submitted to the aforementioned IETF NetVC workgroup. Given that Thor was “years away,” the announcement had the feeling of being more sizzle than steak. Still, that didn’t stop The Broadcast Bridge from breathlessly trumpeting, “Cisco to Nix HEVC with Thor?” At least there was a question mark in the title.
However, the big news came on Aug. 31, when a new consortium called the Alliance for Open Media announced that it’s working on a codec that could be delivered as soon as January 2017. The group includes a number of notable companies, including Google, Amazon, Cisco, Microsoft, Mozilla, Netflix, Amazon, and Intel, and all agreed to forgo royalties on their IP contributions.
The Alliance for Open Media made a big splash, and might have pushed HEVC Advance back toward success.
Significantly, the group consolidated the development of the three major IETF next-gen codecs (VP10, Daala, and Thor), perhaps forever cementing Thor’s place as the world’s shortest-lived video codec. More to the point, this should accelerate the development of the Alliance codec, and assure its selection as the NetVC codec for the IETF.
What are the Alliance codec’s chances? As discussed in this commentary, the standards-obsessed broadcast market will likely stick with HEVC. However, given the rates initially proposed by HEVC Advance, the Alliance codec really looks like a threat to HEVC in browser-based playback.
All this, of course, assumes that the Alliance codec doesn’t infringe upon any HEVC or other patents, which is a real possibility, according to Elvir Causevic and Ed Fish of Black Stone IP, an investment bank that advises operating companies on trading intellectual property and technology assets.
Licensing fees may stall HEVC adoption, while AV1 is nowhere to be seen and VP9 is on the rise. Will 2017 be at eventful as the previous two years?
Netflix compared 5,000 clips from 500 titles in its library using the x264, x265, and libvpx codecs. x265's implementation of HEVC was the clear winner on quality and efficiency, but whether that matters in light of compatibility and licensing issues isn't so obvious.
Speedy adoption of HEVC has been delayed by squabbles over licensing. Here, the developers of x265 propose steps to end the gridlock and move forward.
H.264 still accounts for most video encoding today, but HEVC/H.265 and VP9 are beginning to make noise. What will 2015 bring?