Google announces free WebM video codec as H.264 alternative
At its annual I/O conference, Google has unveiled its plan to release a video codec it acquired as a royalty free alternative to the ISO MPEG's H.264.
Google was joined by Mozilla and Opera as browser vendors that would be including support for the new codec, dubbed WebM. Mozilla's Firefox currently does not support H.264 because doing so would require purchasing a license from the ISO.
WebM is, as the name suggests, targeted at the web and not intended to replace H.264 across the board. As such, the project can be focused on a narrower set of engineering goals. WebM also supports HTML5 delivery of web video, as unlike H.264, it can be used to deliver web-based videos to any modern web browser, not just the commercial products of large companies like Apple's Safari, Google's Chrome, and Microsoft's Internet Explorer.
"The VP8 and WebM specifications as released on May 19th, 2010, are final," Google announced, "and we encourage everyone to use them for developing applications. Google, Mozilla and Opera are all adding WebM support to their browsers and all videos that are 720p or larger uploaded to YouTube after May 19th will be be encoded in WebM as part of its HTML5 experiment."
An open end for On2's proprietary codecs
The technology behind WebM was acquired by Google when it bought On2 Technologies, a company that had developed a series of proprietary codecs dating back into the early 90s. On2's TrueMotion technology began incrementing through a series of new codec generations including VP3, which On2 sold as a plugin for QuickTime, RealPlayer and Windows Media. It later released VP3 as open source in 2002. The free Ogg Theora codec was developed as a fork from the abandoned VP3 code.
Over the past decade, On2 rapidly grew past its old VP3 code to deliver VP4, which it licensed to RealPlayer and AOL; VP5, also used by Real; and VP6, which On2 licensed to China in 2003 for use in its national Enhanced Versatile Disc format as a rival to DVD (which uses the ISO's MPEG-2 video compression standard).
On2 also licensed VP6 to Macromedia in 2005, which made it the primary codec used by Flash Player 8. The successive VP7 codec was licensed to Skype for use in video conferencing. However, by VP7's release in 2005 the world had already began shifting toward open video codecs using a shared patent pool run by the ISO, leaving On2 to pit VP7 against H.264 in a losing battle. Before releasing VP8, On2 was bought up by Google this January.
The move towards open in video codecs
Apple shifted its resources from supporting proprietary codecs like Sorenson (once nearly synonymous with QuickTime in the 90s) toward the iSO's open MPEG technology starting in 2002 with QuickTime 6's support for MPEG-4 and mobile 3GPP video. In 2005, Apple began exclusively pushing the new H.264 with the release of QuickTime 7.
Microsoft had similarly decided to release its own competing Windows Media Video codecs as a published standard, known as VC-1. BluRay licensed both VC-1 and H.264, but Apple's leading position in selling audio and video content in iTunes has nudged global development towards H.264, which is now the most widely used modern codec stretching from mobile video to high definition playback.
The problem with H.264 is that while being an open specification, it's not free. Users have to license the technology from the ISO, which then distributes royalties to the companies who contributed their technology to improve the format. This is a problem for open source projects like Firefox, which don't want to spend money on licensing technology.
The VP-3 based Ogg Theora that Mozilla selected to support instead has serious drawbacks of its own however. While it is free, it's also much older and far less sophisticated than today's H.264 technology, which is still rapidly evolving. Additionally, the path to enhance Theora is littered with patent land mines, as many core video compression technologies are already patented by companies participating with the ISO's MPEG-4, VC-1, or using the patents in their own proprietary codecs.
WebM vs Theora: no contest
Another major problem for Theora is that common silicon doesn't support hardware decompression of Theora-encoded video. Nearly all video chips used in set top boxes or mobile devices now support H.264. This results in better performance and much better battery life for handheld products like smartphones, the iPod touch and iPad.
Google's new free codec, based on On2's unreleased VP8, solves some of the problems Mozilla faced with Theora. The technology is nearly a decade more modern, so it can be taken more seriously than the purely ideological Theora project. It also has the backing of Google, meaning it has both the potential to defend itself from patent attacks and may eventually find support from chip vendors.
Whether Google's new WebM codec will be widely supported on the desktop is not really an issue, as it will be trivial for Google or third parties to offer plugin codec support for QuickTime or Windows Media Player. The remaining problem is finding support for mobile devices such as the iPod, iPhone, and other maker's devices designed to support H.264 playback.
Most mobile and embedded devices can not be upgraded by third parties to support playback of alternative codecs, so even if Google and Mozilla convert lots of proprietary video (such as YouTube's Flash) into free WebM content, today's mobile devices won't be able to play it unless there's also an H.264 alternative available.
However, the new WebM codec should help small publishers deliver their content to desktop users without worrying about licensing costs. And at some point in the future, WebM may be viable on mobile devices if Google can convince chip makers to support the technology in their decoder components. The presence of WebM may also provide competitive pressure on the ISO and its pool of participating technology companies to keep MPEG-4 licensing reasonable for users.
Additionally, WebM further bolsters the move toward HTML5 for audio and video playback on the web, erasing the largest pillar holding up Adobe's Flash platform (even as Adobe rushes to support WebM as well within Flash). At the same time, even with a more modern alternative to Theora, Google may have some difficult in convincing Apple and Microsoft to get on board, given that both have already committed to supporting H.264, and neither see a problem with paying licensing royalties to the ISO.
Google was joined by Mozilla and Opera as browser vendors that would be including support for the new codec, dubbed WebM. Mozilla's Firefox currently does not support H.264 because doing so would require purchasing a license from the ISO.
WebM is, as the name suggests, targeted at the web and not intended to replace H.264 across the board. As such, the project can be focused on a narrower set of engineering goals. WebM also supports HTML5 delivery of web video, as unlike H.264, it can be used to deliver web-based videos to any modern web browser, not just the commercial products of large companies like Apple's Safari, Google's Chrome, and Microsoft's Internet Explorer.
"The VP8 and WebM specifications as released on May 19th, 2010, are final," Google announced, "and we encourage everyone to use them for developing applications. Google, Mozilla and Opera are all adding WebM support to their browsers and all videos that are 720p or larger uploaded to YouTube after May 19th will be be encoded in WebM as part of its HTML5 experiment."
An open end for On2's proprietary codecs
The technology behind WebM was acquired by Google when it bought On2 Technologies, a company that had developed a series of proprietary codecs dating back into the early 90s. On2's TrueMotion technology began incrementing through a series of new codec generations including VP3, which On2 sold as a plugin for QuickTime, RealPlayer and Windows Media. It later released VP3 as open source in 2002. The free Ogg Theora codec was developed as a fork from the abandoned VP3 code.
Over the past decade, On2 rapidly grew past its old VP3 code to deliver VP4, which it licensed to RealPlayer and AOL; VP5, also used by Real; and VP6, which On2 licensed to China in 2003 for use in its national Enhanced Versatile Disc format as a rival to DVD (which uses the ISO's MPEG-2 video compression standard).
On2 also licensed VP6 to Macromedia in 2005, which made it the primary codec used by Flash Player 8. The successive VP7 codec was licensed to Skype for use in video conferencing. However, by VP7's release in 2005 the world had already began shifting toward open video codecs using a shared patent pool run by the ISO, leaving On2 to pit VP7 against H.264 in a losing battle. Before releasing VP8, On2 was bought up by Google this January.
The move towards open in video codecs
Apple shifted its resources from supporting proprietary codecs like Sorenson (once nearly synonymous with QuickTime in the 90s) toward the iSO's open MPEG technology starting in 2002 with QuickTime 6's support for MPEG-4 and mobile 3GPP video. In 2005, Apple began exclusively pushing the new H.264 with the release of QuickTime 7.
Microsoft had similarly decided to release its own competing Windows Media Video codecs as a published standard, known as VC-1. BluRay licensed both VC-1 and H.264, but Apple's leading position in selling audio and video content in iTunes has nudged global development towards H.264, which is now the most widely used modern codec stretching from mobile video to high definition playback.
The problem with H.264 is that while being an open specification, it's not free. Users have to license the technology from the ISO, which then distributes royalties to the companies who contributed their technology to improve the format. This is a problem for open source projects like Firefox, which don't want to spend money on licensing technology.
The VP-3 based Ogg Theora that Mozilla selected to support instead has serious drawbacks of its own however. While it is free, it's also much older and far less sophisticated than today's H.264 technology, which is still rapidly evolving. Additionally, the path to enhance Theora is littered with patent land mines, as many core video compression technologies are already patented by companies participating with the ISO's MPEG-4, VC-1, or using the patents in their own proprietary codecs.
WebM vs Theora: no contest
Another major problem for Theora is that common silicon doesn't support hardware decompression of Theora-encoded video. Nearly all video chips used in set top boxes or mobile devices now support H.264. This results in better performance and much better battery life for handheld products like smartphones, the iPod touch and iPad.
Google's new free codec, based on On2's unreleased VP8, solves some of the problems Mozilla faced with Theora. The technology is nearly a decade more modern, so it can be taken more seriously than the purely ideological Theora project. It also has the backing of Google, meaning it has both the potential to defend itself from patent attacks and may eventually find support from chip vendors.
Whether Google's new WebM codec will be widely supported on the desktop is not really an issue, as it will be trivial for Google or third parties to offer plugin codec support for QuickTime or Windows Media Player. The remaining problem is finding support for mobile devices such as the iPod, iPhone, and other maker's devices designed to support H.264 playback.
Most mobile and embedded devices can not be upgraded by third parties to support playback of alternative codecs, so even if Google and Mozilla convert lots of proprietary video (such as YouTube's Flash) into free WebM content, today's mobile devices won't be able to play it unless there's also an H.264 alternative available.
However, the new WebM codec should help small publishers deliver their content to desktop users without worrying about licensing costs. And at some point in the future, WebM may be viable on mobile devices if Google can convince chip makers to support the technology in their decoder components. The presence of WebM may also provide competitive pressure on the ISO and its pool of participating technology companies to keep MPEG-4 licensing reasonable for users.
Additionally, WebM further bolsters the move toward HTML5 for audio and video playback on the web, erasing the largest pillar holding up Adobe's Flash platform (even as Adobe rushes to support WebM as well within Flash). At the same time, even with a more modern alternative to Theora, Google may have some difficult in convincing Apple and Microsoft to get on board, given that both have already committed to supporting H.264, and neither see a problem with paying licensing royalties to the ISO.
Comments
the more fragmented "web video" format is the better for Flash.
Let's hope the HTML5 standard adds support for all video formats.
Also, Ogg may have some copyright hassles coming at it. So, we'll see.
The licensing issues with MPEG-LA are pretty serious for hobbyist-turned-one-hit-wonder. MPEG-LA stands to be the gateway for all content if h.264becomes the dominant standard.
Unfortunately, it doesn't look like Google can win this one.
I don't think people understand that hardware acceleration isn't an all-or-nothing thing. h264 is virtually fully hardware accelerated because it was specifically targeted, but the fundamentals of how h264 and VP8 work are extremely similar. The most computationally intensive parts of each share the same instructions and algorithms (VLD, IDCT, motion compensation, deblocking, etc).
As they are not completely identical, embedded chips designed to support h264 could easily hardware accelerate the majority of the computations for VP8/WebM, but not all of them. But it's very, very easy to extend those chipsets to accelerate the new instructions VP8 requires going forward -- devices coming out by the end of the year should be 100% hardware accelerated.
For GPU-accelerated videos (desktops & laptops), all that is needed is a driver update. The acceleration is done in "software" on the GPU as it is.
So far I don't think licensing codecs is a problem - isn't H.264 royalty free for another 10 yrs or something like that? Why would we want one more codec - how about making things more simple (KISS principle)?
It's royalty free until 2016 for end-users. Corporations must license it now -- Mozilla would need to pay millions of dollars to license it. Apple, Google, and Microsoft already have paid the money which is why it's not a big deal for them.
But no company seriously supporting "the open web" can support h264 only, which is NOT truly free and is virtually guaranteed in 2016 to cost end users money.
If h264 is ubiquitous and entrenched in 2016, as Apple would like it to be, why would MPEG-LA leave hundreds of millions of dollars sitting on the table? It's only temporarily royalty free for end-users to foster its adoption, then once it's widely adopted they'll be looking to cash in. The companies that are part of MPEG-LA do not develop these technologies out of the goodness of their hearts. The entire purpose of MPEG-LA is to make money from people using their technology.
Remind me what devices have VP8/WebM hardware decoders built-in?
This does not mean that this can not change since it has been less than a day since the opening of VP8 was announced. I'm sure many content providers and hardware producers (e.g. camcorders) will be interested in a codec rivaling H.264 in efficiency which don't come with either a large price tag attached or the looming risk that MPEG-LA might change their mind in 5 years when the current royalty exceptions run out.
one more video format is exactly what Adobe needs.
the more fragmented "web video" format is the better for Flash.
Let's hope the HTML5 standard adds support for all video formats.
Adobe has pledged its support for WebM. The next version of Flash will play WebM videos. Looks like Adobe doesn't care. They're releasing an HTML5 update to Dreamweaver soon. It'll support the HTML5 video tag...and WebM.
Adobe isn't in the codec wars. People keep thinking Flash is a codec...it's not. Flash uses VP6 or h264.
Adding h264 to the HTML5 standard will never, ever happen -- and it shouldn't. It would require any company distributing HTML5 browsers to pay MPEG-LA large sums of money. HTML5 is a true open standard, that cannot happen.
The fact that "mobile devices today can't play it" is a bit of a ridiculous point. Android will support it this year, across all Android devices.
Since over 65% of current Android devices cannot even run the current (2.1 OS), how are "all Android devices" going to be able to do it when another version is released?
The fact that "mobile devices today can't play it" is a bit of a ridiculous point. Android will support it this year, across all Android devices.
I don't think people understand that hardware acceleration isn't an all-or-nothing thing. h264 is virtually fully hardware accelerated because it was specifically targeted, but the fundamentals of how h264 and VP8 work are extremely similar. The most computationally intensive parts of each share the same instructions and algorithms (VLD, IDCT, motion compensation, deblocking, etc).
As they are not completely identical, embedded chips designed to support h264 could easily hardware accelerate the majority of the computations for VP8/WebM, but not all of them. But it's very, very easy to extend those chipsets to accelerate the new instructions VP8 requires going forward -- devices coming out by the end of the year should be 100% hardware accelerated.
For GPU-accelerated videos (desktops & laptops), all that is needed is a driver update. The acceleration is done in "software" on the GPU as it is.
That's a really great and informative post. Does this mean the iPad and the iPhones that have come out or are coming out will not be able to support this WebM format? Did Apple jump the gun?
Don't you mean all "Android devices that are updated with this when it is available" since over 65%% of Android devices cannot run the current (2.1 OS)?
35% are still stuck with version 1.5.
Where do you get your numbers? I'm not sure you understand that 2.0 vs 2.1 is a silly difference, 1.5, 1.6, and 2.0/2.1 are the OS versions right now.
Let's keep this debate on the level, please.
I work at one of the #1 mobile traffic websites in Canada, 75% of our hits come from 2.0/2.1, about 10% come from v1.5 (HTC Dream/Magic). The rest are 1.6 or mods.
Yeah, v1.5/1.6 won't support WebM but within a year the number of devices running 1.5/1.6 will be negligible. 2.0 and 2.1 devices will all upgrade to 2.2 in the near future.
Remind me what devices have VP8/WebM hardware decoders built-in?
Google making VP8 free was a foregone conclusion. The question is why, or more precisely, what is their end game? I think there are two ways we should be looking at here since H.264's decoding on all modern HW will make it the video codec for the inevitable future.
One, will Google be working to push VP8 to be decodable on HW. Two, is this more of a strategic move to get the MPEG Group to make H.264 completely free of charge, thus saving Google a lot of work, time and money supporting dual file types?
Adobe has pledged its support for WebM. The next version of Flash will play WebM videos. Looks like Adobe doesn't care. They're releasing an HTML5 update to Dreamweaver soon. It'll support the HTML5 video tag...and WebM.
Adobe isn't in the codec wars. People keep thinking Flash is a codec...it's not. Flash uses VP6 or h264.
Adding h264 to the HTML5 standard will never, ever happen -- and it shouldn't. It would require any company distributing HTML5 browsers to pay MPEG-LA large sums of money. HTML5 is a true open standard, that cannot happen.
I think you are missing something there.
H264 is not necessarily free, but by any definition it IS an open standard.
Much like HTC paying Microsoft to make Google Android phones, WebM might run into the same situation.
And no I am not saying Flash is a codec, I am just saying the more fragmented the "web video" standard is, the better the future for Flash as a platform.
Where do you get your numbers? I'm not sure you understand that 2.0 vs 2.1 is a silly difference, 1.5, 1.6, and 2.0/2.1 are the OS versions right now.
Let's keep this debate on the level, please.
I work at one of the #1 mobile traffic websites in Canada, 75% of our hits come from 2.0/2.1, about 10% come from v1.5 (HTC Dream/Magic). The rest are 1.6 or mods.
Yeah, v1.5/1.6 won't support WebM but within a year the number of devices running 1.5/1.6 will be negligible. 2.0 and 2.1 devices will all upgrade to 2.2 in the near future.
As of yesterday, his stats look dead-on-balls accurate. I bet if AI listed their web stats Macs and Safari would be considerably higher than the internet as a whole, but one site does not an accurate survey make.
That's a really great and informative post. Does this mean the iPad and the iPhones that have come out or are coming out will not be able to support this WebM format? Did Apple jump the gun?
The iPad and iPhones can support WebM. They won't be fully hardware accelerated, but they can support them with a significant portion hardware accelerated and the rest can run in the NEON component of the chip for the 3GS and later (http://www.arm.com/products/processo...ogies/neon.php -- it's like AltiVec).
I fully expect the iPod/iPhone to never support WebM, though, because Apple is pushing h264 hard as they're part of the patent pool for it. Sites will need to maintain both h264 and WebM versions to hit all clients.
It's not that big of a deal, really. I maintain a site with 22,000 videos and we've got them in 3 different codecs (3GP, VP6/FLV, h264), each of which with at least three different bitrates for quality/bandwidth settings.
Sites like Brightcove abstract this all away from websites as well. You upload in one format, it transcodes and makes renditions for any supported formats. They are supporting WebM also, by the way.
It's royalty free until 2016 for end-users. Corporations must license it now -- Mozilla would need to pay millions of dollars to license it. Apple, Google, and Microsoft already have paid the money which is why it's not a big deal for them.
But no company seriously supporting "the open web" can support h264 only, which is NOT truly free and is virtually guaranteed in 2016 to cost end users money.
If h264 is ubiquitous and entrenched in 2016, as Apple would like it to be, why would MPEG-LA leave hundreds of millions of dollars sitting on the table? It's only temporarily royalty free for end-users to foster its adoption, then once it's widely adopted they'll be looking to cash in. The companies that are part of MPEG-LA do not develop these technologies out of the goodness of their hearts. The entire purpose of MPEG-LA is to make money from people using their technology.
Why H264 can't be a part of Open web? It is by any definition an open standard.
Now if you are talking about "the free web", then yeah it's probably not the best option out there.
It is MPEG-LA, not ISO.
The licensing issues with MPEG-LA are pretty serious for hobbyist-turned-one-hit-wonder. MPEG-LA stands to be the gateway for all content if h.264becomes the dominant standard.
Unfortunately, it doesn't look like Google can win this one.
http://en.wikipedia.org/wiki/H.264/MPEG-4_AVC
H.264/AVC is the latest block-oriented motion-compensation-based codec standard developed by the ITU-T Video Coding Experts Group (VCEG) together with the ISO/IEC Moving Picture Experts Group (MPEG), and it was the product of a partnership effort known as the Joint Video Team (JVT). The ITU-T H.264 standard and the ISO/IEC MPEG-4 AVC standard (formally, ISO/IEC 14496-10 - MPEG-4 Part 10, Advanced Video Coding) are jointly maintained so that they have identical technical content. H.264 is used in such applications as Blu-ray Disc, videos from YouTube and the iTunes Store, DVB broadcast, direct-broadcast satellite television service, cable television services, and real-time videoconferencing.
This might be where the confusion resides with the term ISO and IEC MPEG.
It's an ISO Standard.
As of yesterday, his stats look dead-on-balls accurate. I bet if AI listed their web stats Macs and Safari would be considerably higher than the internet as a whole, but one site does not an accurate survey make.
Yeah, those are global stats from Google. They vary wildly from country to country.
What's interesting is you don't quote the text that accompanies the chart from your source:
Up till last month Google Android 1.5 and 1.6 were always in the lead with versions such as 2.0 and 2.0.1 having a very small usage foot print. For the first time though, Android 1.5/1.6 have been kicked out of the top spot and replaced by Android 2.1. 2.1 accounts for 37.2% according to the chart (larger version below) and that is followed by Android 1.5 and then 1.6.
A few interesting things to note are that some users are still on Android 1.1 which indicates either some small network is not pushing out updates, or some people are just happy the way their phone is.
When Google announce and possibly lunch Android 2.2 (Froyo) this week, it will be interesting to see if this chart skews a lot towards Android 2.2. Google has announced that they want to try stem fragmentation and bring all devices to the same level. Of course there could be some hardware restrictions that would prevent older Android handsets such as the T-Mobile G1 from getting the update, but for all those that can they should get the Android 2.2 update soon.
The trend has already started in replacing the 1.5/1.6 with 2.1, and with 2.2 it'll happen even faster. They're making a concerted effort at reducing the fragmentation. Part of Android's problem was it developed so rapidly with so many partners that it did fragment quite a bit.