Google Opens Up VP8, Launches New Container Format

[V!NCENT]

I believe Phoronix reported it earlier than Slashdot?

Not according to my RSS feeds... /. wednesday en phoronix today...

[droidhacker]

I really don't see the need for the 'webm' format. Matroska is there, why do we need another container with more limitations when Matroska does it so well?

The point of WebM is not to be a general purpose container. It is to be a SPECIFIC container. Basically, ALL WebM files will use the SAME audio and video codecs. This saves you the headache of downloading something just to find that the actual codec used is some proprietary junk that doesn't work. The end result is great simplification for both the producer AS WELL AS the consumer. And when you're using top notch codecs like VP8 and Vorbis, why would you want anything else in it anyways?

[droidhacker]

I couldn't make it to work with the developer preview builds of Firefox. Anyway. There are two licenses, one for the software--BSD-style--and other for the bitstream specification. They also ask for copyright assignment from contributors. It seems their legal department has been working hard to shield Google from patent litigation. This is the text of the bitstream specification license:



So...if a licensee attempts to pursue patent litigation, their rights to WebM granted by this license are terminated. But how does this work with companies, say Apple? Can Google, as a content provider, force Apple and Microsoft to provide out of the box support for WebM and thus be safe from patent litigation? That would be genius.

This is smarter than that.
It does a couple of things;
1) Anyone even TRYING to make trouble over it (successful or not) will lose their rights to use it, which therefore cuts them entirely off of any benefits they could attain from its use.
2) It effectively shuts off any kind of class action litigation since it is unlikely that many groups would want to take the risk of losing VP8. One group might be willing to take the risk, but it is unlikely that they would be able to convince anyone else to.

That means, for example, if, by any chance, h264 uses some google patent covered by this license, and MPEG decides to litigate against google for a patent that google infringes upon, then google retracts everything from mpeg. This would essentially freeze up h264 unless they granted google an equivalent license over whatever patents THEY infringe upon, but google would still deny MPEG the right to THEIR IP, which would basically force h264 into the same license as covers VP8 (since this would be the kind of settlement that google would seek), OR, it would completely kill h264.

An interesting thing you will note about MS's stance on VP8/WebM support: their browser will support it, but only through 3rd party plugins. What this does is it basically frees MS up to litigate against google over patents in this, since they have nothing to lose.

The only other party likely to be hit by this is MPEG, but good luck to them in convincing anyone else to back them up. MPEG will be standing alone and I'm sure that google has SOME kind of patent that they're infringing on. In this world, it is impossible not to be infringing on SOMETHING.
*** that and the fact that you can be assured that some hardware vendors will back up GOOGLE on this: VP8 is more attractive to hardware vendors since there is no licensing involved. So hardware vendors may start pushing back at MPEG saying that they'll DROP MPEG if they don't back off of google. And with google pushing VP8, there will be less and less reason to support MPEG at all!

[droidhacker]

Yeah but what are the chances of including the VP8 codec in a default IE9 installation? (Safari will be able to use VP8 as a QuickTime plugin, too. The million-dollar question is whether the codec will be available out of the box.)

MS specifically said that they will NOT be supporting it directly, but that it will be supported through codec addon, presumably 3rd party.

[droidhacker]

I still prefer H.264 :P

http://x264dev.multimedia.cx/?p=377

That is certainly a long write-up. Doesn't make it right. And given that the guy only had a very short couple of hours to read through the code and write that up before he posted it, I find it EXTREMELY unlikely that his analysis is accurate. I find it much MORE likely that he is a stuck-up h264 lover.

[droidhacker]

A very smart move indeed. A subset of Matroska guaranteeing good streaming, and playback support; ready patches for every meaningful software out there. Youtube transcoded, and just about everyone pitching in to say they support it.

What's Apple's stance? They're the one player so far missing.

With firefox going WebM and definitely NOT h264, pretty much everybody HAS NO CHOICE in the matter. Even MS has fallen in line and announced that it will be available through codec addon. (MS could have gone the "NO WAY" route and blocked it altogether)

[droidhacker]

Ohh man.. WebM, first I was like... YAY.. and then I was like ehhh....

So google bought up on2, who brought us vp3/theora and vp6 all of which are OK, but not super great, everyone was hoping they would open source vp8 and shove it out there saving us from the horrors of the h264/aac proprietary / patent hellworld we live in. My response is best summed up point by point.

1) The name WebM sucks.

2) WebM is based on the MKV / Matroska container, this is FANTASTIC. MKV is a great container, you can shove many audio tracks/subtitles into it with ease and it supports _real_ streaming (unlike the MP4 container), chapters ETC, Divx adopted MKV for Divx7 (Divx seems to be just a name for manufactures to plaster on there portable/dvd/mediacenter/whatever players to mean 'will play random shit you download from the net' but that is besides the point) MKV is the de facto standard for all the HD bootlegs you will find out there on the torrent interweb but I havn't seen it really used outside of that.

3) WebM uses Vorbis audio. This is super FANTASTIC, vorbis is the best lossy audio codec that there is, better than MP3 and AAC. Its also probably actually immune to patents due to its quite different design from other codecs(whatever that could ever mean, since, in my mind software patents are 100% bullshit), to my ears the breakdown from a low vorbis bitrate sounds like Vinyl or FM radio more than the screetchyness of mp3 or AAC. At high bitrates I can't tell the difference between vorbis and lossless. A side note here is that although there are open source AAC encoders, they suck.

4) A note about h264 and x264... x264, the open source h264 encoder, is actually BETTER than any proprietary h264 encoder! ha! But of course, it hurts someones feelings because it violates patents.

5) WebM uses the newly paid for (around 125 million dollars) VP8 codec for video, should be great. But uhm.. Maybe not so great after all. After reading a particularly technical review by an x264 guy, it seems that a) VP8 is pretty close to h264, close enough to be in 'patent danger' whatever that means, and VP8 is essentially a crappier h264. Hrm fuck. Of course, the flaws could be fixed up by the open source community, but the way google is talking the 'spec is final today'. So my understanding is, for there to be real improvements, the bitstream needs to change and therefore won't be exactly VP8 anymore. Somethings got to give here.

http://x264dev.multimedia.cx/?p=377

Summary: Vorbis/MKV rock, WebM is a stupid name, VP8 is much better than theora but wack in many ways. HTML5 video is having one hell of a painful birth, but I suppose thats what happens when you have 4 mothers trying to have the same baby.

Don't put so much faith into some random blog that just happens to be long winded and probably *sounds* over your head. They guy who wrote it did so with the PURPOSE of discrediting VP8 and nothing more. He most definitely DID NOT have enough time to make an educated analysis. He spent a FEW MINUTES browsing the source and specs (which he admits that he is too dumb to understand properly), and wrote a long winded bash against VP8.... all completed within a VERY short time of the data being released by google.

An accurate analysis of code and specs that complex takes WEEKS (if not months).

[droidhacker]

Was there a reason to rename mkv? Well it would be nice when MS would support mkv by default then you don't need to install something else.

The fact that it is different enough to be incompatible.

vp8 is something that nobody really needs, but why not use it as long as Firefox does not support h264...

VP8 differs from h264 in this extremely useful way: linux distros can support it FULLY out of the box, rather than requiring you to add unsupported software manually. This is part of the reason why firefox does not and can not support h264! Because to do so would disqualify it from being packaged with any major linux distro!

[V!NCENT]

[...] to my ears the breakdown from a low vorbis bitrate sounds like Vinyl or FM radio more than the screetchyness of mp3 or AAC. At high bitrates I can't tell the difference between vorbis and lossless. A side note here is that although there are open source AAC encoders, they suck.

OK... you make me want to kill you...

There is no better/worse 'quality' or 'sound' and 'tone' difference between lossless and lossy, but detail. How that is decoded into actual analog sound depends on you audio player. There is only correct and incorrect sound.

The only reason that MP3 is more 'bassy' is that the 'trebble' detail is greatly reduced and because the bass is upped, the treble 'volume' needs to be lowered to not blow up you speaker.

If you actually know what to look for in lossy 'quality' then RIP a track from a CD that has lots and lots of detail to it (amount of synths, bass, shitload of snares and especially some 'background' sounds) and RIP it to 192KB MP3. You'll notice that some tones lost so much detail that they sound like 'waterfalls' or dissapear. Play with the bitrate ripping and you know what the real difference is between lossy and lossless.

If you dislike the hight of tones from the bass/treble, puit an equaliser in place, lol... FAIL.

[pingufunkybeat]

Don't put so much faith into some random blog that just happens to be long winded and probably *sounds* over your head. They guy who wrote it did so with the PURPOSE of discrediting VP8 and nothing more. He most definitely DID NOT have enough time to make an educated analysis. He spent a FEW MINUTES browsing the source and specs (which he admits that he is too dumb to understand properly), and wrote a long winded bash against VP8.... all completed within a VERY short time of the data being released by google.

An accurate analysis of code and specs that complex takes WEEKS (if not months).

That's a bit unfair.

From the post:

Fortunately, it seems I was able to acquire access to the VP8 spec, software, and source a good few days before the official release and so was able to perform a detailed technical analysis in time for the official release.

The "guy who wrote it" is probably the most knowledgeable and relevant person in the world when it comes to h.264. He is one of the main developers of the best H.264 encoder on this planet, bar none. A certain bias is likely, but he does make a number of good points, which are also acknowledged by Greg Maxwell of Xiph. If you don't know who the author of the blog is, you probably shouldn't comment on any video codecs.

I certainly appreciate the comments, even if they put quite a damper on the celebration. But this does not mean that VP8 is not relevant, just that it is not quite as good as the best that h.264 can do, and probably never will be (though the gap will likely close). VP8 is certainly "good enough" for online streaming video, which is exactly what the web needs now, and these are exciting news.

Can't wait for Webkit/Konq/Arora support