Submitted by Earlier this year, Google finally did what many of us hoped it would do: release the VP8 codec as open source. It became part of the WebM project, which combines VP8 video with Vorbis audio in a Matroshka container. The product manager for the WebM project, John Luther, gave an update on the status of the project – and it’s doing great.
WebM has seen a rather incredible adoption rate. It is supported one way or another by all desktop web browsers; natively in Firefox, Opera and Chrome, and through a user-installed codec in Internet Explorer 9 and Safari. It has become part of FFmpeg, runs on x86, ARM and PowerPC, and is supported on Windows, Mac OS X, Linux, and Android. There are codecs for DirectShow, Windows Media Foundation, QuickTime and gstreamer.
There have been 20 new partners, and looking at collection of partners, it’s quite clear that WebM has a very, very bright future ahead of itself. The first VP8 ASIC chips are expected to arrive in the first quarter of next year, delivered by pretty much all the major chip manufacturers.
Furthermore, 80% of the YouTube videos watched every day are already available in WebM. “We are working on transcoding the entire YouTube corpus,” Luther said. Of course, videos barely watched by anyone are much lower on the list of priorities, but eventually, they will be transcoded.
So, what does the future development path of WebM hold? Lots of improvements, mostly. Google is focussing on improving decoding speed, one of the often-cited weaknesses of the current version, leading to an average 28% improvement in decoder speed. They are also working on improving the developer tools.
The title implies that 80% of the videos served ‘daily’ by YouTube are in WebM.
80% are available in WebM (as the article notes) but only a tiny fraction of the videos served daily are served in that format.
(Nothing against WebM. It’s a great standard and a great effort by Google but it’s still really in it’s infancy.)
Good lord, it’s a headline! Headline ambiguity is a standard part of journalism…
The article later clarifies: “available in WebM”.
You missed one word in your statement. Let me rephrase it for you: Headline ambiguity is a standard part of *poor* journalism.
Get your free panties untwisters, right here! Free panties untwisters, get them now!
Thanks. I needed a laugh at bad English slang.
Knickers.
The colloquial Strine saying is “don’t get your knickers in a twist”.
As for panties, apparently according to the kiwis one must not get them in a knot:
http://www.urbandictionary.com/define.php?term=knickers%20in~*~…
Edited 2010-11-12 02:08 UTC
http://www.urbandictionary.com/define.php?term=panties%20in~*~@…
No, its proper Midwestern US slang. Although, you tend to hear:
http://www.urbandictionary.com/define.php?term=panties%20in~*~@…
slightly more often.
Well there you go … you learn something new every day. Unlike Strine, very seldom have I ever heard Midwestern US slang spoken.
Apologies to Thom, I stand corrected.
Edited 2010-11-12 03:34 UTC
Well at least you didn’t snap a jock strap over it.
I’ve always said that you shouldn’t get your panties in a “wad.” Aren’t idiomatic expressions wonderful?
You mean… That kind of WAD ?
http://en.wikipedia.org/wiki/Doom_WAD
opposite side of the marble from the kiwi’s”
“don’t get your panties bunched”
“don’t get your panties in a bunch”
ha.. all over it’s the same idea with different wording.
Here is one for you and tyrione:
http://www.urbandictionary.com/define.php?term=Buzz%20Killingto…
PMBYMHMMFSWGAS about what you think.
http://www.internetslang.com/PMBYMHMMFSWGAS.asp
Edited 2010-11-14 21:49 UTC
Depends on what your goal is
At least it forced me to read the article to find out what it meant.
Edited 2010-11-12 00:32 UTC
That happens when you fall victim to yellow journalism, where the use of ambiguous headlines for the purpose that you describe is known to be common.
Probably… but I guess in my experience, this is the rule, rather than the exception
Yes, sadly mediocrity prevails in many disciplines these days. But IMHO that’s no reason to give up and get used to it.
Kind of like passive aggressiveness, I’d say, mmm, Koki?
Nope, I was simply stating my opinion, which I think I am entitled to, specifically in the context of Urias’ comments.
I can see how you may feel alluded, although my comment was not specifically referring to you or your headline.
Nice try Thom.
To try it out on Youtube you have to enable it:
http://youtube.com/html5
This enables HTML5 video tag usage only for WebM videos, others are still shown with he flash based video player.
Using Youtube completely without the need of a flash runtime however is possible by installing a Script that replaces all calls to the flash video player by a html5 video tag.
I use YouTube.js that can be found here:
http://fabportnawak.free.fr/owb/scripts/
I use it suceessfully with OWB, but it might work with other browsers, too (Opera,Chrome,Firefox+GreaseMonkey).
Please check the documentation of your browser how to install scripts.
The “served” part you added yourself. There is nothing wrong with the text “80% of Daily YouTube Videos Now in WebM”. It is correct to the letter. If Video “Monkey gets shaved” is served today in flash, and is now available in WebM you CAN say “the video ‘Monkey gets shaved’, now in WebM”, because now it is also available in WebM. There is nothing far fetched about it. The title indeed isn’t specific on being about “being served” or “being available” or “of those being served these many are available” but you can easily find out. Would have been a bit too much description in the title, wouldn’t you think?
A little annoyed by the excessive nit-picking. I understood the title just fine on first read.
Note that if you opt into the HTML5 beta for Youtube it defaults to WebM and then falls back to H.264 only if the file is one of the 20% not yet encoded (or you’re on Safari without WebM support) so I’d guess WebM is a big and growing fraction of the videos served as HTML5, though that in turn is probably a tiny slice compared with those served as Flash.
How many YouTube videos are produced, edited and submitted in WebM?
20% of prime-time Internet traffic in the states is Netflix. Netflix has 15 million customers but only 2% – 300,000 – stream video.
H.264 is huge and still growing. WebM needs more than Gooogle and the browswer.
What does this matter?
All new codecs will intially largely have material that is transcoded. It takes a while for anything to be adopted.
For a start, if you are running Windows or Mac OSX, here is a converter for you:
http://www.mirovideoconverter.com/
One can always use such a tool to convert directly to WebM from raw video, it doesn’t HAVE to be h.264 first.
As WebM support exapnds, as it is rapidly doing already, then more and more video will be WebM from whoa to go. From Thom’s article, there are a lot of hardware OEMs who are involved in WebM, and after all, unlike H.264, it will cost OEMs nothing at all in license fees to put WebM in their new camcorder designs instead of H.264.
Enjoy.
Still hits and misses on the quality when trying to convert to webm with this tool (with same webm setting).
converted a 4.2Mb wmv file to 4.1Mb webm file – all good
converted a 192Mb mpeg file to 22Mb webm – poor quality
So webm is fine, but miro converter is not yet.
Not that the parent post is necessarily meant as support for H.264 (although it sure reads as though it is), a number of posters do often submit cheerleading posts to forums such as OSnews in apparent support of h.264. This is something I have never understood:
http://ebb.org/bkuhn/blog/2010/07/07/producing-nothing.html
The development costs of H.264 has been paid for years ago. An revenue intake now is pure cream.
Rent seeking behaviour from vendors of proprietary software is so very clearly NOT in the best interests of the vast, vast majority of people.
Not that the parent post is necessarily meant as support for H.264 (although it sure reads as though it is), a number of posters do often submit cheerleading posts to forums such as OSnews in apparent support of h.264. This is something I have never understood:
It is a strange phenomenon. It mostly seems to come from a vocal minority, fueled by the ultimate selfish desire to use what is deemed top of the line, any other considerations be damned. I often wonder if they would sacrifice their first born, if that would give them access to some fancy gizmo that has been designated the bee’s knees of some use area.
I’d rather opt for 90% of the maximum achievable quality of a certain technology class, if using that doesn’t subjugate the world to draconian licensing restrictions. For some that extra 10% seems to be enough to be willing to sell out their fellow peers.
Even stranger when you consider that there is no case to support a view that h.264 is in any way better than WebM, let alone 10% better.
It is getting more. VP8-decoding DSPs are currently being engineered, as an example by TI (the webM website gives more).
WebM/VP8 does not have to kill H.264 on every single front at the moment. It just has to become the standard format for the web, if we don’t want another Unisys to mess around with us.
What’s outside of the web is another story. VP8’s time in these areas may come later, if it comes at all.
Edited 2010-11-12 18:44 UTC
In case people didn’t read the PDF, the new release of WebM (Aylesbury) improves on decoder performance and encoder quality.
~28% faster decoder speed
7% overall PSNR improvement
6.5% overall SSIM improvement
Next release ‘Bali’, due Q1’2011 will focus on encoder speed but quality will continue to be improved on:
‘We are constantly working on improvements to video quality in the encoder, so after Aylesbury we won’t tie that work to specific named releases.’
http://blog.webmproject.org/2010/10/vp8-codec-sdk-aylesbury-release…
Flash will have to release a player with webm support for webm to become popular.
With flash support, websites will be able to have one media file up and then serve it using flash on browsers that do not support html5/video/webm and webm video on browsers that do.
Currently, webservers will have to have a separate file for this format and no practical thinking person will do that.
Despite Adobe saying they would support WebM when Google announced it, lack of Adobe Flash support for WebM to date is one of the only disappointing things about WebM progress so far:
http://gigaom.com/video/80-of-youtube-videos-now-available-in-webm/
Edited 2010-11-12 02:21 UTC
Interesting.
HTML5 and Flash have many overlapping features, video streaming being one of them. No wonder Adobe is careful about choosing allies.
Does it mean Adobe considers HTML5+WebM a tougher competitor than HTML5+h.264, though? The latter is now more widespread but the former has lower (zero) marginal cost.
Another possibility is that MPEG LA has some special exclusive license agreements with some of their major partners.
Turn the question around, and look at it another way.
Wouldn’t Flash+WebM be more acceptable (because of the zero marginal costs) than Flash+H.264? Wouldn’t it be possible for Adobe to ship Flash with both WebM and H.264 support for now? Wouldn’t it be better for Adobe if WebM took off and Adobe could drop H.264 support later on and hence no longer have to pay royalties to MPEG LA for every copy of Flash?
The one and only feature where Flash has a significant advantage over HTML5 right now is adaptive bitrate streaming. Adobe would want to keep ahead of HTML5 if they want to stay relevant, and lack of WebM support isn’t helping the case for Flash right now.
It won’t take all that long for open source to work out how to provide adaptive bitrate streaming for WebM and then incorporate it into HTML5. Adobe have only a very short timeframe in which to try to keep Flash relevant.
http://blog.gingertech.net/2010/10/09/adaptive-http-streaming-for-o…
http://planet.xiph.org/
Edited 2010-11-12 04:46 UTC
In contrast to Flash it has smaller CPU requirements and I have a much better experience. Gnash could support VP8 in order to help convert more content.
Fedora14x86_64 + Atom D510MO
How can I find if I have the latest VP8 implementation?
Not that the parent post is necessarily meant as support for H.264 (although it sure reads as though it is), a number of posters do often submit cheerleading posts to forums such as OSnews in apparent support of h.264. This is something I have never understood
H.264 has a life beyond the web.
In studio production. In cable, sattelite and broadcast distribution. In industrial and security video. In home video.
There are about 30 H.264 licensors and 900 H.264 licensees. Most are Asian. They include global giants in manufacturing like JVC, Mitsubishi, Panasonic, Samsung. Sony. Toshiba.
Google is big. But it ain’t that big.
For a company the size of Mitsubishi, the cost of licensing H.264 is no more significant than the price of a single diet soda sold from the vending machine downstairs.
None of which makes H.264 a good choice for a web video codec.
Whereas the fact that H.264 is not royalty free means that it does not comply with patent policy for web standards.
http://www.w3.org/Consortium/Patent-Policy-20040205/
WebM overcomes the deficiencies of h.264 as a web codec.
Therefore, the question must remain, why are there cheerleaders for h.264?
Cheerleaders, certainly, but what I see more often than that is the blind general acceptance of H.264. The fact that Slashdot could report on burgeoning HTML5 video support and turn a blind eye to the issue of H.264 is just stunning. It seemed as if OSnews was the only news site actually _questioning_ Apple’s use of H.264 video in HTML5, we were reporting on it in various forms for weeks. It even lead to this letter from us: http://www.osnews.com/story/23235/A_Statement_Regarding_OSNews_Bias
Because here I am waiting for a WebM plugin for Quicktime or a WebM encoder for Mac OS X as to allow me to encode my uncompressed videos into WebM videos. I’d use it if the WebM advocates provide it – the operating system vendors have already decided, if you want the WebM presence to improve then provide plugins, encoders and so forth for those of us on such platforms. Those of us who are happy with h264 do so because you’ve failed to provide an easy and accessible alternative to h264.
mencoder.
The End.
Give me a nice front end and a binary from a reliable source without needing to use something like ‘darwin ports’ then I might consider that an option.
Ah, you have all the power of SUS3 Unix at your fingertips, but you’d rather push buttons for limited options.
‘kay.
Not surprised, just… facts.
I remember, years ago a kawai that didn’t shun the power of the command line. Sometimes I wonder where he went.
If your doing encoding what is the benefit of a pretty set of buttons you are going to set once for any particular batch of data and then statically stare at when the data encodes? It’s like watching paint dry with a progress bar.
Trying to create an interactive and dynamic interface for a non interactive process is silly. Even in the case of having multiple work flow outputs there is no benefit to a shiny interface that let’s you select a profile over a text file flag in a command line. Its tantamount to saying there is no point to using gcc without a pretty gui.
For any given video to transcode people do not constantly change resolutions, bit rates, post processing options per scene. In plain terms, editing is not done with an encoder. Encoders, shockingly, encode and then ship the output somewhere else. If anything the benefits of ‘ | ‘ far out weigh pretty buttons.
Edited 2010-11-12 17:30 UTC
Double post
Edited 2010-11-12 17:31 UTC
FFMPEG is available via the Fink project.
Try avidemux. It now supports VP8 in MKV containers and there is a Mac binary for it.
http://fixounet.free.fr/avidemux/download.html
WebM for Quicktime, alpha release announcement in August:
http://groups.google.com/a/webmproject.org/group/webm-discuss/msg/a…
August, I hope they get one done in the next 6 months because if they’re trying to get people to use it over h264 they’re taking their time in achieving such a goal.
http://www.mirovideoconverter.com/
Thank you for the link – what is the encoding quality of VP8 like when set on the highest encoding quality settings? (I’m ok with waiting a few minutes longer).
Kaiwai wrote:
Thank you for the link – what is the encoding quality of VP8 like when set on the highest encoding quality settings? (I’m ok with waiting a few minutes longer)
I think the quality is good, not as good as when I use x264 with really high settings (I haven’t tried the newest release yet though) but definately good enough for web content.
The encoder speed on the other hand needs alot of work, obviously it is currently compared to x264 which is chock-full of hand-optimized assembly and pretty much state of the art in encoder optimization but this is a good thing in my book as long as people don’t expect it to reach reach that level of maturity over night. Google will need to improve greatly in this area, and it seems they know this since the next release due in Q1 2011 will focus on encoder performance.
I believe there’s every reason for h264 and Webm to coexist as there’s certainly a need for a free-to-implement video codec on the web in order to level the playing field, and on the other hand for those who has the capital and want to provide the very best quality for those who are prepared to pay for it I say h264 is and most likely will remain the best choice.
Not because there is a technical barrier in making VP8 as good as h264 but rather that some of the techniques needed to reach that level of quality are likely heavily patented.
Of course the losers in this is the MPEGLA, since they were well on their way into cornering web video through their h264 codec and stifling any competition with their patent pool. But the rest of us are better off since we can still pay to use h264 if we want or go the free route and use webm.
Edited 2010-11-13 01:40 UTC
Well, it is a very hot topic.
Some people like Dark Shikari say that VP8 is noticeably worse than H.264 on highest quality, using some specific frames of HD public domain videos that are known to be hard to encode.
Other must look very carefully in order to notice the artifacts, and think that in the moving video no one will notice the difference.
Myself, I belong to the second category, and what’s more I think that at low bitrates, VP8’s blur looks much, much better than H.264’s blocks as an artifact.
So… I’d say that VP8 is worse at encoding small details, but has a better look overall. But my advice would be “try it yourself and see how well it goes”.
Edited 2010-11-13 08:34 UTC
That seems to me to be a reasonable summary … where there is a shortage of bandwidth … high mtion at low bitrates … h.264 produces relatively sharp frames which exhibit artefacts, while WebM produces blur, especially the parts of the picture in motion.
One might argue that the human eye actually sees rapid motion as blur anyway. I don’t believe the eye sees sharper images plus little blocks that aren’t there.
Having said that … it must be said that judgements like these are very subjective, and different people might prefer either one over the other.
If you want to see how good VP8 can get for web use, see the 720p version of this video with HTML5 enabled.
http://www.youtube.com/watch?v=rLxQiI8c1Bs
In the way, I’ve found out what’s the problem with this announcement from Google : most of the videos that are accessible in WebM/VP8 format are only accessible in their 360p version. And transcoding + low bitrate + low res means absolutely awful quality for a lot of videos I’ve seen.
On the other hand, contrary to what happens when using the flash player, HTML5 video playback doesn’t randomly freezes in the middle of a video and never comes back, forcing the user to seek somewhere else before the player starts buffering again (in Opera 10.63). Switching resolution during playback does seem a bit buggy, though : I’ve seen the video player crash because of this several times.
Edited 2010-11-13 09:14 UTC
http://code.google.com/p/webm/downloads/detail?name=webmquicktime-0…
http://www.mirovideoconverter.com/
Both of these will work on a Mac, I believe.
Bear in mid the WebM encoder is still very slow, as people have been concentrating on decoder performance first.
The only reason why Google has been able to encode 80% of its YouTube video to WebM when the WebM encoder performance is still so slow is that Google has masses of computing power available to it.
So, the software in links above should work, but don’t expect to be impressed just yet by the encoding speed.
For other platforms, other resources are here:
http://code.google.com/p/webm/downloads/list
Enjoy.
kaiwai :
WebM QuickTime Component 0.1.0
Uploaded: Oct 05, 2010
They got it done for you in a little over a month, just about 6 weeks.
Edited 2010-11-13 07:47 UTC
Interesting that you appear to be throwing in the towel for h264 regarding it being a web codec and then defending it in a lot of areas where WebM is likely to encroach in the coming years.
Even more interesting is you telling us how cheap it is for companies to license. A per-device royalty for each device is a pretty significant cost for any company, let me tell you.
The h264 bubble that MPEGLA had to keep intact has burst and there’ll be no containing it.
Is there a fast and easy way to create a webm video, e.g. a tool like ffmpeg2theora?
There are patches available to ffmpeg, which may have been merged by now, which allow it to encode to vp8+webm.
Google has made some unfortunate marketing choices here. What is webm? It’s VP8+Vorbis+A container. What is the container called? If there’s some name for it other than webm I do not know it. So I can say “ffmpeg supports encoding vp8” but I cannot say “ffmpeg supports webm” because it’s not clear what’s being supported here. Audio? Video? Container? All three?
Webm is the container. It is directly designed to house vp8 + vorbis but I can’t see any reason why it couldn’t house other codecs (maybe ffmpeg already supports webm as a container for other codecs?).
But as it stands, when people speak of Webm they will most likely refer to it as containing VP8+vorbis so I don’t see any reason for confusion here.
??
No. It’s mkv.
It’s the same container.
Same thing.
Exact same file format/internal structure. The extension is different, indicating expected contents.
The extension is not all that’s different. The webm matroska-based container is a subset of mkv.
…Yeah, it’s limited to containing VP8 and Vorbis.
That’s part of the standard for WebM.
However, the container format itself is MKV.
An MKV with VP8 and Vorbis data in it renamed to .webm or whatever is the same thing.
But calling it webm makes it alot easier to know what codecs are in question than if they had stuck with mkv. And it’s not like the matroska guys mind, in fact they seem very happy with webm:
http://www.matroska.org/news/webm-matroska.html
Not contesting that. Just stating that it _is_ the same container format.
Except that it’s a subset. You know what “subset” means, right? All webm files are matroska files, but not all matroska files are webm files. So, it is not possible to simply change the extension of a .mkv that has vorbis audio and vp8 video. It is only possible if it conforms to the subset of matroska specified by webm.
…Which is basically ‘contains only VP8 and Vorbis audio, and conforms to MKV specifications’.
So… it’s the same fucking thing.
mkvtools 0.3.3 is the reference tool for dealing with containers!!!
IT’S THE SAME FILE FORMAT.
WebM is a container and a marketing effort. This is a problem because the marketing effort means it’s hard to know when someone is talking about *just* the container.
It’s just a marketing effort, really.
It’s a standard that takes something that could already be done, and… standardising.
Is there a fast and easy way to create a VP8+Vorbis+MKV?
Been using pitivi to transcode for youtube with good results.
pitivi from
https://launchpad.net/~gstreamer-developers/+archive/ppa
?
I thought this new-fangled WebM thing looked interesting. I tried WebM on YouTube. My Pentium 4 3.0 GHz w/HT was dropping mondo frames on a 360p video.
It may be html5 video that is to blame, and not WebM. I have had heavy frameskip in my browser when I used an Athlon 3000+ on Linux some times ago, and now reading the previously mentioned video on Windows 7 + Opera 10.63 eats up half of one core of my core i5 M.
Try to save the video and directly read it in a video player which reads WebM (I think VLC does), and see if performance is better.
Edited 2010-11-13 19:43 UTC
Sounds like a browser issue, on a single core Atom N270 w/ GMA900 graphics on ubuntu 10.10 WebM playback is very good.