Slashdot Mirror
MPEG-LA Considering Patent Pool For VP8/WebM
An anonymous reader writes "Well, that didn't take long. Larry Horn, CEO of MPEG-LA, the consortium that controls the AVC/H.264 video standard, says the group is looking at creating a patent pool license for VP8 and WebM, Google's new open source, royalty-free HTML5 video format... So much for a Web video standard unencumbered by patent issues." We talked about VP8/WebM a couple of days ago when Google open sourced it. Reader Stoobalou points out another late-night email from Steve Jobs, who was asked to comment on VP8 vs. H.264. Jobs laconically sent a pointer to the technical analysis we linked before, where the poster says "VP8 copies way too much from H.264 for anyone sane to be comfortable with it, no matter whose word is behind the claim of being patent-free."
Without that guarantee of return, these companies would have never bothered to invent VHS, CD, MPEG in the first place.
Maybe these companies wouldn't have. Does this mean that these technologies would not have been invented anyway ? I don't think so. MP3 was invented partially thanks to public funds (Fraunhoffer is a half-public R&D institute) and most of the patented "innovations" of private R&D labs are often base on public research publication. What my experience showed me is that usually, in "R&D" the "R" is often made by public labs and the "D" by private companies. Therefore, it should be lawful to reuse the "R" part. In the context of software, that often means the core algorithm, exactly what is concerned by these patents. The 'D' part is well covered by copyright.
The Wise adapts himself to the world. The Fool adapts the world to himself. Therefore, all progress depends on the Fool.
IANAL, but, wouldn't that make a good case for estoppel? Saying that there were well known, visible implementations of this technology in the field for years, and the rights holder waited until commercial viability and adoption to give economic incentive to litigate. Seems like the definition of estoppel to me (in my NAL understanding).
My Babylon
Quite so. Google now owns every On2 patent, and except for the ones in VP3 (Theora), which On2 already basically relinquished, Google has exclusive control over those licenses.
Google are a search company, and they are very, very good at it. They routinely search millions of terabytes of data in seconds. They have a patent search database: http://www.google.com/patents - and presumably have international data as well as USPTO data but don't yet make that public.
It is reasonable to assume that before they spent $100 million buying On2 for the sole purpose of being able to know a video codec is clean, they could have done an exhaustive patent search. Exhaustive. As in every extant US patent prior to the date of On2's patents. They could quite reasonably have done so. They have the resources to do so, all of the data, a very good search algorithm, and a lot of smart people. Frankly, nobody else, including MPEG-LA or any of its members, could have reasonably done so, and would scoff at whether it was even possible. They very probably did exactly that—that's my guess, anyway. They won't guarantee you, because they're not in the business of selling you insurance (and of course, MPEG-LA are, hence the "nice codec you got here, shame if something were to happen to it" tone).
But I'd bet a fair sum that VP8 was specifically designed by On2 to skirt, but carefully avoid, every MPEG-LA patent (which is why it uses a standard DCT, and why it doesn't have B-frames). (Remember you have to infringe all of the claims of a patent to be infringing.) If On2 had infringed an MPEG-LA patent, MPEG-LA would have been on the attack years ago.
Unless, of course, they know an ugly truth. Maybe H.264 infringes on an On2 patent, and MPEG-LA don't have a licence for it.
Google are now in a position to simply say that they will license those patents to anyone under the license they just published, of course—which simply prohibits any meaningful retaliatory strikes against WebM—and not under any other license. If you want to use H.264, fine, no problem by them, although you'd need an MPEG-LA licence as well. As long as you don't attack WebM or anything in it.
Presto: mutually assured patent destruction defuses the situation, and VP8 gets just as good an assurance as MPEG-LA have.
Also, you know, they kind of own YouTube. You know, the site that serves more video than everything else combined. If your browser, say, Safari, stops working with YouTube, people are not going to view it as a problem with YouTube, especially given Google Chrome is built on the same core but is significantly faster.
VP8 doesn't look too bad. It isn't quite x264 quality but it's way better than Theora, it's vastly faster to decode than H.264 is in software, hardware stuff is already on fab and possible with many of the same parts and extensions (any Android phone will be able to do it, and will in 2.2 I think). The encoder is kind of poor, but will get much better after a bunch of open-source tinkering I think. It prefers blur to block, which is much better when you're talking about a small screen too; really, H.264 can be overly grainy and blocky and in many cases tries to be too sharp. VP8 is good enough for the web, and it's open, and though it isn't going to be evil, it does has a powerful gorilla protecting it now.
Also, y'know, by the by, so does Matroska and Vorbis.
Win-win. Bring it on.
But I'd bet a fair sum that VP8 was specifically designed by On2 to skirt, but carefully avoid, every MPEG-LA patent
This. There's absolutely no other reason why the spec would "borrow" so much from h.264 and yet come out worse... unless they started with h.264 and removed everything that was patented.
If I have been able to see further than others, it is because I bought a pair of binoculars.
Yeah, but even MPEG-2 is better than MPEG-2.
That's not a joke or a typo. The codec world has moved on quite considerably since the release of MPEG-1, and development of MPEG-2 encoders has resulted in stunning improvements in the last few years, in part because of the requirements of ATSC, and in part because of the improvements in processor technology.
I've been wondering for a while if the right approach for the whole codec mess is to wait until MPEG-1 is truly free and clear, and adopt that. You may laugh, but try encoding something in MPEG-1 with ffmpeg, using large (> 100) GOP sizes, and high numbers of B-frames (16+.)* On a normal high performance computer in 2010, the speed of compression is too low to be practical, but the results are excellent, even at relatively low bitrates. 4-6Mbps is more than enough for high quality 720p24, in most cases.
For that reason, I think the Internet video codec debate will be over sooner than people think. The real work has to be done on the encoder side, improving the capabilities of encoders for older formats that'll be patent free soon. But if you look at the bigger picture, MPEG-1 video (and MPEG-1 layers 1/2 audio) will soon be free (some claim they already are), bandwidth is improving, CPU power is improving, While a superb MPEG-1 encoder will never be as good as a superb H.264 encoder, the necessity of one over the other will diminish in time.
You are not alone. This is not normal. None of this is normal.
Yes, the problem is that there are some really bad patent thickets out there, and codecs are one of the worst patent thickets (also in terms of enforcement).