A Blog From Behind the Trenches

nice

edit: but opera isn't linked yet

Available Now
Chromium tip/nightly trunk build beginning May 19, 2010
Mozilla Firefox nightly trunk build beginning May 19, 2010

Coming Soon
Opera beta
Google Chrome Early Access Release Channel beginning May 24

btw: opera isn't mentioned there at all -> http://www.youtube.com/html5

opā, that's cool

O yeah ! Nice !

http://files.myopera.com/Tamil/Smilies/Happy.gif -

Can't get to http://webmproject.blogspot.com/ have to log into Blogger to view

So good!

Whoa this will make some difference! Will it be effective on mobile devices like iphone and when is it getting mainstream with opera (10.5*?)?

I was right! When Google show the world the Open VP8, I knew this is would happen

Operaaaa

Hell yes. Hopefully with all this muscle H.264 can stay away from the Web. There's still problems left to solve, but with Google's influence perhaps hardware decoders could be produced. What's remaining is a place for end users to actually embed the videos on their web pages in this format. It's violating the terms of use of YouTube to embed a WebM or H.264 video on your website isn't it? Afterall, having a place to upload all those large videos to in the first place was what made video take off. Something similar would be necessary for standards-based video.

Now what about easily accessible tools to produce this format? Manually patching ffmpeg is okay for someone like me, but I'm sure many others wouldn't bother with that kind of thing.

Fantastic, death to H.264 and it's patents.

I am looking forward to this stuff hitting the major tech blogs and seeing Microsoft and Apple's responses.

Fun times ahead.

Adobe will also include VP8 in flash so legacy browsers can have flash (IE9, Safari).

Benchmarks are welcomed.
Does it have advantages over AVC, besides the cost for manufacturers?

Commentary from a video compression expert including some benchmarking:
http://x264dev.multimedia.cx/?p=377

Seems slightly worse, but competitive and I'm looking forward to future speed and quality improvements. I am a little worried about patent infringement though

Originally posted by mgillespie:

Originally posted by mgillespie:

cost for manufacturers

Patent-Free.

Actually the same reason.
By advantages I mean engineering solutions. How good is VP8?

Originally posted by Astrophizz:

Commentary from a video compression expert including some benchmarking:
http://x264dev.multimedia.cx/?p=377

tnx

Originally posted by mgillespie:

Patent-Free.

+1

freezes on 720p quality

This is awesome! *_____*

Originally posted by xErath:

Adobe will also include VP8 in flash so legacy browsers can have flash (IE9, Safari).

http://windowsteamblog.com/windows/b/bloggingwindows/archive/2010/05/19/another-follow-up-on-html5-video-in-ie9.aspx

EDIT: It seems I was late

There is no right click context menu on WebM video - but perhpas its mentioned in some article here - I haven't read all of them yet.

Huzzah! Sometimes google really is not evil.

Incidentally, why was the linux version of the labs build not provided as a tarball? That way everyone could have used it.

Drat, Dillon got here before me with http://x264dev.multimedia.cx/?p=377 (I was busy reading the whole thing and going through the discussion below.)

Better container format (though ripped off and re-branded), terrible documentation and questionable future. I'm checking out the Labs build now, but feel that VP8 will be taking all of us on a long and bumpy ride.

PS: The browser should be able to play any correctly coded file via context menu or drag-n-drop, I first noticed this lack with .ogv files when Theora support landed. You have better AVI support than HTML5 support.

PPS: Thanks much for the decision to donate your GStreamer porting work.

Another Follow-up on HTML5 Video in IE9
http://windowsteamblog.com/windows/b/bloggingwindows/archive/2010/05/19/another-follow-up-on-html5-video-in-ie9.aspx

Dangit are we all posting the same three links today?

So I took it for a spin on my Atom N280-based netbook. Results for YouTube are as follows:

1) No fullscreen yet.

2) Slick interface otherwise, mostly "just works".

3) Looks about the same as Flash, often sounds worse. (YouTube's fault?)

4) More CPU usage and frame-tearing than Flash version. (less efficient)
a) NO I DO NOT HAVE ACCELERATED FLASH

5) Seems to generate larger cache entries than h.264 (less compression)

6) Seems to use more memory, but this is less certain.

VERDICT: I think they've managed the "good enough" quality level if they can X) improve audio quality, Y) lower system requirements, Z) survive the patent litmus.

Oh yeah! Can't wait for WebM support on my Android device!

I can't make it work on Ubuntu 64, I get:
"Your browser does not currently recognize any of the video formats available.
Click here to visit our frequently asked questions about HTML5 video."

Originally posted by Guille:

I can't make it work on Ubuntu 64, I get:
"Your browser does not currently recognize any of the video formats available.
Click here to visit our frequently asked questions about HTML5 video."

Don't install it, run it 'in place', i.e.

$ tar xf opera-10.54-21867-webm.i386.linux.tar.bz2
$ opera-10.54-21867-webm.i386.linux/opera &

It seems we have a small bug that results in the install script not copying the WebM files into the correct location.

If you really want to use your main profile (not at all recommended but it is your choice), change that last line to:

$ opera-10.54-21867-webm.i386.linux/opera -pd ~/.opera &

P.S. Since installing does not work right now (at least as far as WebM support, which is the point of the labs release) we have removed the install script from the online *nix packages.

Originally posted by mgillespie:

Have Opera messed up again with their marketing and missed the boat? What still amazes me, is Opera wonder why they have such a small marketshare, when they let things like this happen, and do such a poor job at marketing the best browser on the market.

Oh, quit it with the crazy rants already. It doesn't take more than a tiny non-issue, and you start ranting about all sorts of completely irrelevant crap. Save your knee-jerk reactions for something that actually matters. That you are mindlessly parroting the market share myth just proves how deluded you are.

Originally posted by BS-Harou:

There is no right click context menu on WebM video

That has got nothing whatsoever to do with WebM. It's how Google chose to add it to YouTube.

Originally posted by Astrophizz:

Commentary from a video compression expert including some benchmarking:
http://x264dev.multimedia.cx/?p=377

More like commentary from someone who is financially, emotionally and creatively invested in H.264. Heavily invested. Take his biased rants with a grain of salt.

Originally posted by hellspork:

VERDICT:

It's pretty insane to make a verdict based on alpha quality builds intended for early testing of the feature.

Originally posted by prd3:

That has got nothing whatsoever to do with WebM. It's how Google chose to add it to YouTube.

no, it's about how Opera implemented it) ... there is default context menu for Theora too)

Originally posted by BS-Harou:

no, it's about how Opera implemented it) ... there is default context menu for Theory too)

Wrong.

Originally posted by prd3:

Wrong.

well the whole point of HTML5 video is that it is right in the browser (not is some plugin) so its browsers choice what is in the context menu...

Originally posted by prd3:

More like commentary from someone who is financially, emotionally and creatively invested in H.264. Heavily invested. Take his biased rants with a grain of salt.

He isn't emotionally or creatively invested in H.264. In fact he's stated he might work on WebM encoding. He's said he would work on Theora except that the code is apparently a mess. In fact many of the new improvements coming to Theora are a result of his suggestions to the Xiph team. Sure he might be biased, but as it stands that post of his is the most informed article on the internet about the technical merits of WebM. Please stop attempting to discredit knowledgeable people based on your gut reactions.

Originally posted by BS-Harou:

well the whole point of HTML5 video is that it is right in the browser (not is some plugin) so its browsers choice what is in the context menu...

It IS right in the browser. It's just added to the page in a different way.

Originally posted by Astrophizz:

He isn't emotionally or creatively invested in H.264.

He IS invested in H.264 since he has spent enormous amounts of time and resources on x264. Claiming otherwise is simply insane. He has a history of attacking anyone who doesn't bow to H.264's (according to him) excellence. He has a history of distortion, including this latest hit piece against WebM.

I'm wondering what people think about Google not indemnifying people who use WebM?

Google backs open codec against patent trolls

Anyone who wants to sue over WebM will have to face Google's army of lawyers anyway.

BTW, I'm wondering what people think about MPEG-LA not indemnifying people who use H264?

That link doesn't say anything about Google protecting users of VP8 from litigation. On the other hand I read some stuff and apparently not indemnifying end users is par for the course for this kind of thing.

Originally posted by prd3:

He has a history of distortion, including this latest hit piece against WebM.

But he is quite right - h264 is a superior video codec today.

Originally posted by Aux:

But he is quite right - h264 is a superior video codec today.

Not the baseline, which is what people are restricted to if they want their all-important hardware support.

Originally posted by prd3:

Not the baseline, which is what people are restricted to if they want their all-important hardware support.

Yep, but x264 dev noted that in his article - VP8 stands very close to baseline. Personally I think baseline gives sharper image, but also some strange artifacts. So they both have their own cons and pros.

Yeah that's the difference as it stands right now. x264 by default opts for psychovisual optimizations that increase the visual detail at the cost of some artifacting and slightly lower PSNR whereas VP8 as it stands just optimizes for PSNR. Google has stated that they will add "HVS" (human visual system IIRC) optimizations so I would look for psychovisual optimizations in the future.

PRD3: I was comparing the currently available viewing experience using the currently available video sources. I was comparing Google's x264 implementation against Google's VP8 implementation using publicly available playback software. I am not especially pleased, and the webM roadmap shows that the dev team wants to fix some of these issues ASAP.

The matter of skipping-around frame quality is a serious matter, especially in terms of CPU load and bitstream integrity. The current output of the VP8 encoder will be tough for mobile devices to swallow, especially with decoding only done in software. Hardware-decoding professionals have stated that VP8 will be difficult to implement due to an overly precise filtering step that is better suited for general-purpose processors. This may also limit the utility of hardware acceleration using modern graphics cards and their very simple stream processors.