YouTube and VP9: A Made-for-Press-Release Event
The recent news that YouTube will demonstrate 4K video encoded with VP9 at CES with hardware support from a number of chip and TV vendors has all the earmarks of a made-for-press release event; all froth, no substance. That’s OK; press release writers have to eat too. But before you lose faith in H.265/HEVC, you should consider the following facts.
VP8 Had Hardware Support, Too—Supposedly
This isn’t the first time Google has tried to establish an open and royalty-free alternative to a commercial video format. Google’s VP8 video codec, which the company released in 2010, was supposed to become the default format for plugin-free video streaming and real-time communications, but those plans were thwarted by a lack of hardware support and fierce opposition from some companies with vested interest in established commercial video formats
Poor, poor, Google, right? All that fierce opposition. What GigaOM doesn’t mention is that Google announced extensive hardware support when they launched VP8/WebM. Here’s a bit from a GigaOM article on May 19, 2010.
The project has support from a variety of hardware makers, including ARM, AMD, NVIDIA, Texas Instruments, Broadcom and Qualcomm. On the publisher side, it’s getting support from Brightcove, Ooyala, Encoding.com, Kaltura, Zencoder and others.
Hey, it’s many of the same names! As you may recall, Adobe also announced that it would support WebM in the Flash Player. Just like Google’s own promise to remove H.264 support from Chrome, very little of this support, if any, ever materialized. So when it comes to chip vendors like NVIDIA and Broadcom, let’s wait and see.
They’re Selling TVs, Not Streaming
With TV vendors like LG, Panasonic and Sony joining the announcement, it’s clear that the focus is 4K TV, not VP9 in general-purpose streaming. These companies need 4K to succeed because TV sales are contracting for the second year in a row—they’d join a marketing initiative if Richard Simmons or Dennis Rodman claimed they were going to support 4K. Unfortunately, according to ABI Research, the 4K TV market isn’t going to take off anytime soon.
Senior analyst Michael Inouye commented, “Despite a very limited installed base there have already been a number of 4K trials from broadcasters, pay TV operators, and satellite operators. While many point to the 2014 FIFA World Cup in Brazil and Sochi Winter Olympics as highlights for 4K, these events will have a minimal impact on 4K adoption – it’s simply too early.”
Practice Director Sam Rosen added, “Unlike 3D, which required awkward glasses, 4K has the legs to become an industry norm. This isn’t a sprint, however, and it will take time for the necessary infrastructure, installed base of devices, and content to come together before 4K becomes an integral part of how the typical TV household consumes video content. We expect this could start to happen as early as 2018 in some regions (emphasis supplied). In the meantime, many consumers will have 4K panels without 4K content, or 4K game consoles without a 4K display, and will claim a superior 4K experience even though the technical merits are not quantifiable.”
TV manufacturers will join any marketing initiative that makes it look like 4K is coming sooner than it really is. As explained below, this has very little impact on the general-purpose streaming market.
What About VP9 Playback on Computers?
While the OTT market within which Sony, Panasonic, and LG play is increasingly important, it’s not mainstream streaming. If Adobe announced support for VP9 playback in the Flash Player, or Microsoft and Safari announced VP9 support in their respective browsers, I would sit up and take notice. Until they do, it’s hard to take any of this noise seriously for general-purpose streaming.
Given the patent issues discussed below, and in particular, Apple’s intransigence for supporting proprietary technologies (other than its own), it’s almost impossible to imagine that VP9 will gain iOS or Safari support anytime soon, if ever. Think about it. If you asked 100 video producers today whether they cared more about iOS decode or 4K on YouTube, what percentage would say 4K on YouTube?
What About Mobile Decode?
I recently reported that the most concerning issue for video publishers with HEVC was not the potential for content-related royalties, but the lack of battery-efficient software decode. I haven’t seen a lot of data about how much CPU it takes to decode VP9 on computers much less on mobile. Until VP9 can efficiently play on mobile devices, it’s a non-starter for most streaming producers, even it just for Android delivery.
VP9 vs. HEVC Quality
It’s hard to find an unbiased opinion comparing the quality of HEVC to that of VP9. A report by members of the IEEE, who arguably aren’t unbiased because HEVC is a standard, concluded,
According to the experimental results, the coding efficiency of VP9 was shown to be inferior to both H.264/MPEG/AVC and H.265/MPEG HEVC with an average bit rate overhead at the same objective quality of 8.4% and 79.4%, respectively. Also, it was shown that the VP9 encoding times are larger by a factor of more than 100 compared to those of the x264 encoder.
In other tests, Google’s own engineers found VP9 to be 7% behind HEVC back in 2011; though closer, they were still behind. Going forward, there are literally dozens of hardware groups working to improve HEVC quality, but only one, at Google, working to improve VP9. That’s the major disadvantage of a proprietary codec. That said, this didn’t really hinder VP8, which was very close to the quality of H.264. But overall, if you had to bet on one horse leading the quality race, it would be HEVC, not VP9.
Lingering Patent Issues
As I talked about here, Nokia has sued Google for VP8/9 patent infringement in 50 jurisdictions. While Google did win one ruling in Germany, here’s what the authoritative Foss Patents blog had to say.
In my report on the second VP8 trial I explained that the make-it-or-break-it issue in this case is a single claim construction question that could be viewed differently on appeal or in different jurisdictions where the same European patent or any of its non-European equivalents might be asserted. In that post I also said that I've seen the order to reopen the proceedings in Nokia's first VP8 case after a finding of a likely infringement . Recently Nokia brought its third patent assertion against Google's VP8 video codec (part of the WebM Project) in an ITC complaint.
Nokia provided the following statement on the VP8 ruling:
"Nokia respectfully disagrees with the decision of the court and we will consider our options. This is only one case among 50 patents originally asserted against HTC, and Nokia has recently filed further cases. To date, HTC has been found to infringe Nokia patents and an injunction is already in effect in Germany as a result of those decisions. HTC needs to end its unauthorized use of Nokia's proprietary innovations."
In summary, Nokia won a case before a Dutch appeals court, narrowly lost a VP8-related one in Germany (which it may appeal and/or pursue in other jurisdictions), and one other Nokia case was stayed, which benefits HTC for the time being but could still result in an injunction if Nokia defends the patent before the Federal Patent Court of Germany.
It’s hard to say how real Nokia assertions are in the marketplace; you’d have to assume that it was considered and discounted by all the chip vendors who are considering supporting VP9. At the very least, these patent claims will serve as Apple’s, and perhaps Microsoft’s, justification not to support VP9.
Where’s the Bandwidth?
The most intelligent estimate of how much bandwidth it will take to deliver 4K TV is between 12-15 Mbps. As one comment on the GigaOM article stated, “Youtube is buffering enough with 1080p already, they should fix that before introducing 4k.” True that.
While GigaOM did quote Google as saying, “the use of the codec won’t just help YouTube to deliver higher resolutions at reasonable bitrates, but also reduce the amount of data necessary to stream regular HD videos by about half. This will help YouTube to improve video delivery and do away with buffering, said Varela: ‘By 2015, you’ll be surprised every time you see that spinning wheel.’”
Nice words, but that assumes general-purpose playback in the PC and mobile space, which as mentioned above, isn’t here. According to NetMarketShare today, Internet Explorer enjoys over 51% of the desktop market share, with Safari at around 7%. How many publishers will adapt a technology that can’t deliver to about 60% of the market? About as many as adopted WebM.
Figure 1. Browser market share as of January 6, 2014.
Mobile share is similar; with iOS enjoying over 54%. While iOS’ share is inevitably shrinking, there’s no question that it’s far and away number one mobile brand in terms of mindshare for video producers. That’s not only because Apple is a preeminent brand, but because they’ve done a better job making it easy to stream to their devices, while Android remains a stinking mess to deliver to. And that’s something at VP9 files on YouTube won’t make any easier.
Figure 2. Mobile share
Again, while YouTube can and will deploy whichever codecs it desires, most websites won’t support a technology that they can’t deliver to over half the potential viewers, which is precisely why very few websites supported WebM. Until we see some action by Adobe, Apple, or Microsoft, it’s hard to see any web video producer getting excited about VP9. And, as you’ll see in the next section, YouTube hasn’t ruled out supported HEVC, so it can get the same savings (or more) supporting a standards-based codec that will ultimately play back everywhere.
If you view the recent spate of VP9/YouTube-related articles through the lens of TV vendors needing to sell 4K TVs, you get the proper perspective. Remember, Google made plenty of noise with WebM/VP8 and YouTube, and the fact remains that YouTube is one of the few publishers that actually supported WebM.
In the video codec space, standards enjoy a tremendous advantage over proprietary technologies. While it’s far too early to count out VP9, the recent made-for-press-release event will have little impact in the general-purpose streaming space.
What is interesting is whether Google will attempt to use YouTube as leverage to force Microsoft and Apple, and perhaps Adobe, to support VP9. After all, if 4K video from YouTube was only watchable in Chrome and FireFox, it could impact browser market share to some degree. In this regard, Google has made it clear that VP9 isn’t an either-or decision. “This certainly isn’t a war of the video codecs,” a YouTube spokesperson said according to GigaOm. The spokesperson then added that this was just a first announcement around 4K for YouTube, leaving open the possibility that YouTube could add H.265 support as well.
So, we’ll have to see if Google does play the YouTube card to get Apple and Microsoft to buy into VP9. Until, then, the YouTube/VP9/4K noise-fest is, like the man says, “nothing to see here, folks, just keep it moving.”
VP9 is the open-source codec from Google, and provides a royalty-free alternative to HEVC. It's more efficient than H.264, and while it's less efficient than HEVC, it compares well on quality.
At this year's CES, some TV vendors asserted that HEVC was done for and VP9 would win out. What's happened with VP9 since then?
Which codec delivers better image quality? Which is more compatible? And what about Daala, the spoiler codec currently being developed from scratch?
Companies and Suppliers Mentioned