Nokia refuses to license patents it says are needed to use Google's video technology, sullying Google's earlier patent deal. But WebRTC could still spread VP8 widely, lowering Web video costs for startups and schools.
Justin Uberti, a WebRTC leader and VP8 advocate at Google, speaking at Google IO 2012.
A Google patent-licensing deal two weeks ago dramatically improved the fortunes of its VP8 video technology, but Nokia has added a barricade to what has already been an arduous road to adoption.
VP8 is a codec -- technology to encode and decode video or audio data for compact storage and efficient network streaming. Despite passionate debates about VP8 vs. the incumbent codec, H.264, most people need never care about video codecs.
But as video becomes ever more deeply embedded in the Net -- TV entertainment, chatting with friends, videoconferences for business, online schooling for children -- the video codec issue becomes ever more important. At stake in the current debate is whether H.264 and its big-business licensing terms will prevail, or whether there also is room for an open-source, free-to-use alternative that could give an edge to cash-strapped startups, schools, and self-publishers.
Spreading VP8 Google and allies including Mozilla want to build VP8 into the workings of the Net as a royalty-free technology. The current focus for the debate is WebRTC, a technology for Web-based chat that members of the Internet Engineering Task Force are standardizing.
Two weeks ago, Google announced a VP8 patent deal that could have helped persuade IETF members that VP8 support should be a mandatory part of WebRTC.
WebRTC has a lot of momentum -- it's even got
its own conference -- so it could conceivably carry VP8 all around the Net. That would start with browsers, but then could extend into smartphone processors, video-streaming sites, set-top boxes, and video cameras.
Google's patent deal was with
MPEG LA, which licenses pools of video patents used for various standards, and it cleared some of the clouds of intellectual-property risk that loomed over VP8.
But Nokia last week told IETF members that
it's got patents that bear on VP8. And in discussions about the related WebRTC, an IETF standard for browser-based audio and video chat, efforts so far have failed to produce consensus that VP8 should be required as mandatory to implement (MTI).
Google has steadily pushed for years its aspirations for royalty-free video on the Web, including through its $123 million acquisition of VP8 developer On2 Technologies. WebRTC gives Google, Mozilla, and others a second chance to build VP8 into the workings of the Web -- something they failed to do years earlier when hammering out how to build video directly into the standard for Web pages themselves, HTML. WebRTC is built into Chrome and Firefox and about to arrive in Chrome for
Android, and developer adoption would help spread VP8 far and wide.
Patents are a significant problem to VP8 and its use in WebRTC, but not the only barrier. The biggest issue, arguably, is simply that the industry so far has largely coalesced around H.264, aka AVC, and it looks likely to move smoothly to its successor, H.265 aka HEVC.
The technology is well-understood and broadly supported. With H.264, makers of Web browsers, video cameras, mobile-phone processors, and DVDs pay royalties to MPEG LA which licenses a pool of patents for the codec.
Computing industry players are accustomed to H.264 royalty payments -- indeed, Cisco would like H.264 to be the mandatory WebRTC codec. But that price is a big deal to smaller players.
"Speaking as a developer working for a company with much less money than yours," Lorenzo Miniero of Web-based videoconferencing company
Meetecho told Cisco in a
mailing list post, "if the IETF selects H.264 as the MTI codec, this will make it even more dramatically difficult and expensive (if not impossible) for pretty much everybody else to use WebRTC."
Also, patent-encumbered technology is fundamentally at odds with open-source software such as Mozilla's Firefox, and therefore can't tap into the energy and cooperation of open-source programming.
Google works to steadily improve the performance and image quality of its software for encoding and decoding VP8 video. It does the same with freely available technology that lets chipmakers build VP8 support into hardware. This graph shows the latest hardware encoder, Foxtail, producing significantly higher quality than the first-generation Anthill version at the same network transmission rate. Alternatively, it offers the same quality at a lower bitrate. Foxtail also beats out Duclair; the decoding software.
(Credit: Google/WebM)
"Having open-source codecs will allow for faster acceptance and faster improvements to the codec by the development community," said Paul Greenlea, a programmer and strategist at development shop
Fresh Tilled Soil, which just published a
demonstration of WebRTC-based chat using Chrome running on a
Google Nexus 7 and on a MacBook Air
.
Google works hard to improve VP8 as part of the higher-level WebM project, which bundles VP8 and the Vorbis audio codec for streaming video. Periodic releases of freely available VP8 encoders and decoders technology, usable either as software libraries or for building into video hardware, show steady improvements in image quality at a given network transmission data rate.
Google's MPEG LA patent deal
Google's VP8 deal with MPEG LA, announced earlier this month, gave Google "a license to techniques that may be essential to VP8" held by 11 patent holders.
One snag, though: In 2011, 12 organizations came forward to say they had VP8-essential patents when MPEG LA said it was investigating the formation of a VP8 patent pool.
Nokia, apparently the 12th company, isn't playing along. "The declaration says that Nokia is not prepared to license the listed patents for RFC 6386 under any terms,"
Nokia's Markus Isomaki said on the IETF's WebRTC mailing list.
RFC 6386 is a Google description of VP8 at IETF.
In addition to trying to
coax the tech industry to adopt VP8 as mandatory in WebRTC, Google is trying to standardize VP8 itself. In January, it brought VP8 to the
International Organization for Standardization, submitting VP8 to the
ISO working group known as the Motion Picture Expert Group (MPEG). That's the very group that produced H.264 and its brand-new sequel,
HEVC/H.265.
VP8 allies had hoped to convince WebRTC standards group members that VP8 should be a mandatory-to-implement codec. At a November 2012 meeting, the WebRTC group was set to see if there was consensus on the matter -- but
Google abruptly yanked the item off the agenda and pushed it back to the next IETF meeting in March.
The MPEG LA deal arrived shortly before the March meeting was set to begin, though. Among those who bridled to see the agenda including a WebRTC codec discussion were representatives from Apple and Microsoft, two companies that are big H.264 fans; they wanted more time to digest the MPEG LA deal and more information about what companies are actually involved in it.
"The decision rests on (at least) late-breaking news, and unavailable information, and I am opposed to any move to get a formal decision for a mandatory video codec at the meeting," Apple's David Singer said in a
mailing list message. "There is some very important data (notably, the names of the 11 and the resulting license) which is promised to be available in only a few weeks. Let's not decide in haste (and possibly repent at leisure)," he
added.
And Microsoft's
Matthew Kaufman said Google's announcement "was posted too late to be of any value for this meeting."
Ultimately, the IETF group again postponed WebRTC codec issue.
Catching the mobile wave
But that was just agenda wrangling. In another high-traffic discussion item, Internet giant Cisco championed H.264 as the mandatory WebRTC codec, arguing that VP8 would hold the standard back. It came from a
post by Jonathan Rosenberg, chief technology officer of Cisco's cloud collaboration group:
I would like to suggest that, at this time, adoption of VP8 as MTI [mandatory-to-implement codec] will slow down adoption of WebRTC by turning off developers that would otherwise embrace it if H.264 were selected...
The reality is [that] today's Internet communications systems are built on H.264. And unless the developer cares only about living within the island of the web browser, a VP8 based solution will simply not meet their requirements...
If IETF selects VP8 as the MTI codec, this will make it dramatically more difficult and expensive for us to use WebRTC.
Google employees offered rebuttals. Google's
Justin Uberti objected to the idea of the Web as an island, referring indirectly to the number of Firefox, Chrome, and Chrome for Android browsers that support WebRTC: "I think it is worthwhile to point out that within weeks, there will be 2B+ [more than 2 billion] deployed WebRTC endpoints, across desktop and mobile, that support VP8. Surely that dwarfs the total number of deployed H.264 endpoints by several orders of magnitude."
Added Harald Alvestrand, a Google VP8 author, in a
mailing list message, "When we have a growing market, and are intending to do a change, we always have a choice between early and late change. The later we leave the change, the more devices there will be out there with 'before-the-change' technology deployed. If we think the switch to VP8 makes sense at any time in the future, it's less expensive to start the switch now than to start the switch later."
And the hardware support is coming so mobile devices won't use so much power decoding VP8 video.
Alvestrand included a VP8 image quality test that describes the state of hardware support:
More than a dozen chip manufacturers have announced chips with 1080p VP8 support, including Samsung (Exynos 5), NVIDIA (Tegra 3), Marvell (Armada 1500), Broadcom (BCM28150), Texas Instruments (OMAP54xx), Freescale (i.MX 6), STEricsson (NovaThor L9540), LG Electronics, Hisilicon (K3v2), Rockchip (RK2918, RK3066), Nufront (NS115), Ziilabs (ZMS40) and Allwinner (A10). Clear majority of leading mobile chipsets in 2013 will contain VP8 hardware support.
VP8 will catch on
Ultimately, VP8 will find a place, said Karl Dahlin, vice president of business development at videoconferencing hardware company
AVer. H.264 is widely used for corporate viceoconferencing chat rooms, but things will change with the spread of mobile videoconferencing.
"The rush of consumer-based videoconferencing today and new applications of WebRTC will use VP8, and mobile videoconferencing users will eclipse [traditional video chat] rooms very quickly," with 100 times as much usage, Dahlin predicted.
Although it's easier to focus development and testing resources when there's only one standard, that's not going to be the future of video, he said.
The end result of all this wrestling will be quite simple, he said: "The infrastructure for video should support both."
Correction, 10:52 a.m. PT: The caption beneath the graph incorrectly identified the top performing encoder shown in the graph itself. That encoder is Google's Foxtail VP8 encoder.
