Google to open-source next-generation VP8 video compression codec

[Andrew Smith]

This is huge. Think of VP8 as being the codec that we would have possibly moved to after finishing with h.264.

See Google to Open-source VP8 for HTML5 Video

Andrew

[Jenn Kramer]

Dunno if this will happen or not but one of the companies I work with licenses VP6 from them for flash video encoding and after the Google purchase the price dropped from around $1,500 per 4 core server per year to $200. I wouldn't be at all surprised if Google open sourced VP8, assuming it's fairly patent un-encumbered, which is the big question.

[Andrew Smith]

The good thing about Google purchasing the company this developed the codec is that they now own the technology and hence any patents that were related to it.

This clears the path for a common and superior codec (VP8) for all browsers as part of a roll-out of HTML5s ability to play video through the use of a 'video' tag and not having to use a complex script invoked Flash wrapper. Due to the pure open source nature of the Mozilla Firefox project, they are unable to include the proprietary newsexyfangled h.264 codec. (Browsers will need the codec software code to decode and play out the video content directly instead of the flash wrapper doing it.)

Of course, people can still use h.264 (or other codecs) inside a Flash wrapper. That said, the HTML5 way of doing things is the future and Flash video will remain as a fall-back for browsers that aren't HTML5 savvy yet.

On the other hand, there are some very well qualified individuals who feel that Flash player 10.1 will kill HTML5.

Andrew

[John Vincent]

It only means one thing - another potentially confusing fight for consumers caught in the middle of a corporate pissing match.

As a producer of content, I'm interested, but not at the expense of ease of use for the masses.

H264 enjoys a big head start and is a fine way to deliver a project. Even google might not be able to change that.

john

[Andrew Smith]

Actually, I doubt that consumers would even notice the difference. At most they might notice a lot more customisation / skinning of the video player interface inside the web page.

For web site developers and related video professionals, it gives us an extra option. Particularly handy when the royalties are finally payable on h.264 (currently deferred until 2015). I'd personally rather re-encode to VP8 than have to start paying royalties on h.264 if I was a large scale provider.

It will probably be another 5 years before we can tell the shape of how things are going to work out.

Andrew

[Jack Zhang]

VP8 may be the codec of choice for distribution, but H.264 will still be a key acquisition format due to formats like AVCHD and AVC-I.

This would also allow Firefox to finally have a more broader support of HTML5 video since at present, HTML5 is primarily H.264, something the Mozilla foundation want no part of.

People saying that Flash 10.1 will kill HTML5 are talking in terms of GPU acceleration. It won't be long after VP8 goes mainstream that we see hardware decoders for that codec built into video cards and soon also be GPU accelerated.

[Tim Polster]

Interesting.

How long until we need to start knowing this stuff? 6 months, 2 years?

The last thing I want to be is the first guy on the block with a new video format on my website that nobody but a few tech folks can play.

[Andrew Smith]

At the moment, HTML5 is still being finalised as a standard so web video usage won't be refactoring until (a) the standard is finalised, and (b) there is sufficient support in the dominant web browsers.

In terms of encoding now, it's not worth it as (so far as I can tell) the Flash video player doesn't support the decoding of the VP8 compression format yet. Support for h.264 was only added in December 2007)

I'd estimate that the earliest use of the VP8 format would be as a HTML5 implementation compression codec once browsers are supporting it. You would still have a regular Flash video as a fallback for browsers not recognising the HTML5 code. (Note: Two files (one for each codec) would be required on a web server during this intermediary stage.)

In the meantime it's a very intersting thing to follow. What Google have just done for us is a Good Thing™ and kudos to them!

Andrew

[Andrew Smith]

BTW, here's a reason to get seriously excited ....

"Leading H.264 implementations require as much as twice the data to deliver the same quality video as On2 VP8. In addition, the On2 VP8 bitstream requires fewer processing cycles to decode.

Multi-core processing is one of the most significant developments in computer technology. On2 VP8 is designed to take full advantage of multi-core systems and can efficiently use up to 64 processor cores simultaneously."

From an info page on On2 VP8. More details on that page including a link to a full media kit.

Andrew

[Dave Haynie]

Originally Posted by John Vincent

It only means one thing - another potentially confusing fight for consumers caught in the middle of a corporate pissing match.

Not so much, I don't think.

Originally Posted by John Vincent

H264 enjoys a big head start and is a fine way to deliver a project. Even google might not be able to change that.

Google has the magic of vertical integration. They control the CODEC, at least one Web browser, and the go-to destination for video distribution. If VP8 actually looks good at half the bandwidth of H.264 (this is really stated for low bitrate stuff, I don't think they're worrying about replacing H.264 on cameras or Blu-Ray), Google could save a bundle streaming VP8 rather than H.264. And they do all the encoding, too. So they have complete control.

If this does save Google money, it'll probably save others money as well. So they'll move in this direction. Could be that just the higher bitrate stuff moves to VP8 (again, assuming the claims are realized)... no VP8, you can watch in "HQ" H.264 or standard Flash/VP6.

This does kind of settle the HTML5 <VIDEO> tag issue... well, as well as anything. You have one CODEC that's offering lower bitrates, high quality, and open source. What's not to like?

Devices.. that's the real problem. Most devices have H.264 acceleration, and simply won't play either Ogg Theora or VP8, unless those algorithms can be run on the existing video acceleration hardware. Otherwise, you suck CPU power and kill batteries for a bad viewing experience. But of course, you don't need HD on most portable devices.

This kind of illustrates the one remaining problem: some of these guys, like Mozilla, still want to implement the video decoder in the browser itself, rather than use the OS's own video API. This is a bad idea.. you're unlikely to get the same level of optimizations, which can mean bad video quality and more CPU power spent on the decoding. Every major OS, even the smartphone OSs, have their own video CODEC APIs now. The web browser really needs to get out of the OS business as much as possible, and just concentrate on doing its main job.

[Andy Wilkinson]

More info on this today

Google launches open WebM web video format based on VP8 (update: hardware partners and Microsoft statement) -- Engadget

[Andrew Smith]

Wow. Who would have thought that h264 could possibly be surpassed so soon.

Now all I need is an updated version of Matrox's CompressHD and I'll be in a very happy place!

Andrew

[Dave Haynie]

Originally Posted by Andy Wilkinson

Google launches open WebM web video format based on VP8 (update: hardware partners and Microsoft statement) -- Engadget

This is great! Google was widely expected to open source VP8, but they seem to have done it very, very right. They've dumped the brain-damaged Ogg container, instead supporting Matroska, which is a modern structured container format, on-par at least with MP4.

And they have the hardware guys behind it. The big deal of making this a successful alternative to H.264 is support on devices. At least at current web video resolutions, you don't absolutely need video acceleration on PCs... but they do have AMD and nVidia on that list, the two guys pretty much responsible for video acceleration on PCs, so that's good, anyway. But then there are handheld devices...and this list is a who's who of ARM makers: Freescale, nVidia, TI, Qualcomm, ARM themselves, and MIPS... the other widely used embedded architecture.

Actually, the one thing that surprised me a little... Microsoft. Not that they would support it out-of-the-box, but after all the talk of "H.264-only" in IE9, they're saying you have to install the VP8/WebM CODECs and then IE9 will support it. That's actually suggesting they're even "doing it right" in IE9. The Web brower really has no business being a video decoder... it should be passing that job off to the OS. Microsoft can correctly claim "H.264 only" out of the box if they do this, since they now include H.264 with Windows, but none of the other CODECs of interest. But this is much better than their locking the choice into IE9, which they certainly could have.

Given their supporters, it's easy to believe that this could be distributed with Chrome and other browsers, with nVidia and ATi drivers, etc. You're probably going to have it, even before you need it.

As for "surpassing H.264"... well, I haven't see VP8 in action yet.. On2 didn't publicly release it before Google bought them. They claim 40-60% more CODEC efficiency on very low bitrate material... that's not exactly saying it will surpass H.264 in all things. Then again, the whole industry has been pretty regularly wrong at knowing the limits of these things. MPEG-3 was supposed to become the CODEC for high definition video, but MPEG-2 did the job well enough to not worry about developing MPEG-3, MPEG-4 was the system for the web, with H.264 optimized for phones and other small things... but now taking over HD as well.

[Andy Wilkinson]

Here's a comparison between H.264 and On2 VP8....

On2 video - Making Video Possible

[Dave Haynie]

One of the x264 coders comments on VP8:
Diary Of An x264 Developer The first in-depth technical analysis of VP8

He doesn't seem to be a fan, and suggests there may be patent issues. Of course, one has presume Google's army of lawyers reviewed everything before they did this whole WebM thing... stay tuned!

You can get DirectShow filters and other binaries here:
Downloads - webm - Project Hosting on Google Code