Move over, Theora and H.264 - One more HTML Video option appears

May 19, 2010 1:50 PM

As rumored and requested, Google has released VP8 as a royalty free video codec, complete with patent license. Add in Vorbis for audio and Matroska as the container format, and you have WebM, soon to be supported in Chrome, Firefox and Opera. YouTube support is already here.

Apparently, IE9 support will be available through installing a VP8 codec.

Jason Garrett-Glaser reviewed VP8 and the released encoder/decoder, and found it to be inferior to H.264 but better than Theora. He also feels it's worrying similar to H.264 and finds it "a patent time-bomb waiting to happen."

posted by dragoon (57 comments total)

Is some algorithm geek working on a storage format for videos thats tweaked so that they can be easily transcoded to H.264 / Theora / MP-4? Is that even possible? I'm wondering if sites like youtube are going to have to store multiple versions of every popular video?
posted by BrotherCaine at 2:03 PM on May 19, 2010

That review is pretty damning. Have there been any responses to it from Google or On2?
posted by Fraxas at 2:05 PM on May 19, 2010 [1 favorite]

I'm interested to see when and if Safari adds support for this.
posted by reductiondesign at 2:05 PM on May 19, 2010

It's important to note that the review is written by an h.264 developer. He obviously doesn't want to call VP8 the savior of the internet.
posted by DoublePlus at 2:10 PM on May 19, 2010 [4 favorites]

Apple is pretty well invested in H.264 as a standard across all of their platforms, so I'm sure they'll drag their heels on proper support in Safari as long as they possibly can.
posted by Oktober at 2:10 PM on May 19, 2010

Google has a lot of very, very smart, math-oriented people working for it. One analyst has gone so far as to describe it as "math club." I think the On2 acquisition is far more important than the release of VP8. This pairs resource-heavy (money, PhDs) Google with a company who knows video — not perfectly, obviously, just look at that review — and ought to produce some interesting results.

VP9 is what I would look for. I anticipate the resultant codec to draw away from any problematic patents in H.264. The spec would be cleaner and more testable. The compression and optimization are tertiary, at best. Nobody wants a repeat of the Unisys GIF incident, and this time, the stakes are that much higher. Transcoding from one lossy codec to another is Bad News. At least with GIFs, many people had original files or could turn them into PNGs with a little love.

This is my hope, at least. With YouTube, which is, let's be honest, a vast bulk of video data, Google has, if not a very long lever arm, one wallopingly-sized rock to put the end of it.
posted by adipocere at 2:14 PM on May 19, 2010 [4 favorites]

On further thought, what I'd like to see happen is Apple purchasing full rights, patents, everything they need to own H.264, and releasing it for free in response.

Win-win.

/pipe dream
posted by reductiondesign at 2:14 PM on May 19, 2010 [1 favorite]

Who indemnifies users of H.264?
posted by BrotherCaine at 2:15 PM on May 19, 2010

Who indemnifies users of H.264?

MPEG-LA
posted by Zed at 2:19 PM on May 19, 2010

I'm wondering if sites like youtube are going to have to store multiple versions of every popular video?

Disk space is cheaper than processing power, so I expect most video sites will just store multiple copies.

Who indemnifies users of H.264?

Well, if you got your implementation from someone who paid MPEG LA then at least you're clear of those patents. Beyond that, it doesn't appear that anyone else has stepped forward to sue h.264 implementers, so that enters into the calculated risk equation.

But of course no one is ever going to sue users of any codec. It would be preposterously expensive, time-consuming, and unlikely to net anything. No, far more likely would be to sue h.264 implementers for a reasonable royalty. Or, much more efficiently, to simply join the MPEG LA patent pool.
posted by jedicus at 2:20 PM on May 19, 2010

That review is pretty damning. Have there been any responses to it from Google or On2?

That review is only from today.
posted by smackfu at 2:21 PM on May 19, 2010

FYI I'm giving a talk on this in more detail in a breakout session at 4:15 here at I/O. I can come back and answer more questions after that (pretty busy at the moment).

(If any of you are here, stop by!)
posted by wildcrdj at 2:23 PM on May 19, 2010 [4 favorites]

Oops, I answered a different question than the one actually asked. H.264 is paid for by the people who license the codec, including Apple and Adobe. If I understand the current state of affairs correctly, encoding to H.264 and publishing on the web is guaranteed free for not-for-profit use for some years, as is having and using an H.264 codec for playback, but it's possible these could change eventually. (But that's the end of 2015 or 2016.)
posted by Zed at 2:24 PM on May 19, 2010

Not to go too far off topic, but how does x264 fit into the whole licensing thing?
posted by bstreep at 2:29 PM on May 19, 2010

I've tested it out on Youtube using the nightly Firefox release - the encoder or decoder definitely has some kinks, as one of my favorite Youtube videos has recurring blockiness show up out of nowhere, but in general it looks pretty good.

The list of supporters on the WebM blog is very impressive - ARM and Broadcom are talking about hardware support, so hopefully mobile support will come sooner than later on platforms beyond Android.
posted by dragoon at 2:36 PM on May 19, 2010

Not to go too far off topic, but how does x264 fit into the whole licensing thing?

It doesn't. That is, it's an unlicensed implementation of h.264. Note, however, that the h.264 patent owners have not sued the developers or the users of x264, including commercial users.

The list of supporters on the WebM blog is very impressive

Interestingly, it has no overlap with the h.264 patent licensors.
posted by jedicus at 2:40 PM on May 19, 2010 [1 favorite]

Great to see this actually happen. I sincerely hope Apple and Microsoft support this open standard for the benefit of their customers as apparently that's what motivates them.
posted by juiceCake at 2:44 PM on May 19, 2010

It's important to note that the review is written by an h.264 developer.

Just to be clear, he's an x264 developer, an open source unlicensed encoder/decoder, and has nothing to do with the corporate interests that could be involved.
posted by floam at 2:45 PM on May 19, 2010 [2 favorites]

Interestingly, it has no overlap with the h.264 patent licensors.

And unfortunately, those licensors carry a lot more financial weight. A lot.
posted by reductiondesign at 2:45 PM on May 19, 2010

Great to see this actually happen. I sincerely hope Apple and Microsoft support this open standard for the benefit of their customers as apparently that's what motivates them.

if it's in any way technically inferior to H.264, I'm not sure how that'd be helping their users. The users are already enjoying H.264 content that exists all over the place in large quantities that's readily accessible on their current devices and computers and software. The only users I see being screwed over right now are users of Mozilla's stuff, who are being hurt by someone else's fear of software patents.
posted by floam at 2:48 PM on May 19, 2010

So here we have a strong, free, open codec available for inclusion in HTML5 and for video playback on mobile devices. A viable alternative to flash video without the patent nightmares of h264.

Think Apple will put their money where their mouth is? Or will the fact that this is from Google be too damning?
posted by kafziel at 2:49 PM on May 19, 2010 [1 favorite]

The only way to get Steve Jobs on board is if there's a way for apple to force a "no-porn" flag on it. :P

(seriously, you guys don't actually think apple would ever deign to lower themselves to an open codec, do you?)
posted by symbioid at 2:52 PM on May 19, 2010

Or, let's put it another way: If the quality of the codec was as good or better than H.264, do you think Jobs would include it? Despite his whole schtick about "the user" and "quality" I have a feeling he'd still go H.264 over VP8...
posted by symbioid at 2:54 PM on May 19, 2010

if it's in any way technically inferior to H.264, I'm not sure how that'd be helping their users

Can you play MP3's on iPods and Macs? It's an inferior format to FLAC and AAC and yet it's supported.

Mozilla users aren't screwed over by Mozilla, Mozilla and others devoted to avoiding a repeat of the GIF fiasco are being screwed over by those with money enough to pay for and damn support of rival, free codecs.

Besides, I thought things improve and are not locked or frozen in time in regard to quality. I see arguments like hardware support for h.264. At one time, theatres nationwide lacked hardware support for sound in projectors. Then sound came into cinema and after all these years, finally, theatres everywhere support it, despite the shift in hardware. I've even heard of hard drive based projectors in use these days...
posted by juiceCake at 2:58 PM on May 19, 2010

My new codec requires infinite bandwidth, completely preventing digital distribution.

I am fully funded by the MPAA and RIAA.
posted by blue_beetle at 3:10 PM on May 19, 2010 [4 favorites]

Can you play MP3's on iPods and Macs?

On the other hand, the range of video formats you can play on iPod Touch/iPhone is extremely limited.
posted by Artw at 3:59 PM on May 19, 2010

Also from emerging from Google today: Google Font API, Google Chrome Web Store

Nothing mobile related - I wonder if that is day 2?
posted by Artw at 4:03 PM on May 19, 2010

As Artw points out, the iPod is limited to certain video codecs, which, to be honest, is a pain in the ass. Every video I want to load has to get sent through a program so I can load it.The PS3 is less limited, but it does not play .mkv. Then again, very little plays .mkv (VLC and...?).

God, I just want a happy pleasant format that will play on the devices I own without needing to sift through crappy re-encoding software. While some might give out a big huzzah to new formats, I just sigh, and figure I'll need to find a program to switch that over, too.
posted by Ghidorah at 4:11 PM on May 19, 2010

Google Font API, huh? Typekit can't be too pleased about that.
posted by reductiondesign at 4:17 PM on May 19, 2010

Can you play MP3's on iPods and Macs? It's an inferior format to FLAC and AAC and yet it's supported.

It's kind of moot because MP3 is entirely ubiquitous. If real audio was as popular as MP3s they might support that too. My point was only that nobody is screwed due to not having support for it. That could entirely change if it ever became popular. Apple supporting it would be awesome, and a Good Thing, but it wouldn't have anything to do with benefiting their consumers.
posted by floam at 4:18 PM on May 19, 2010

Google Font API, huh? Typekit can't be too pleased about that.

Apparently they are in on it.
posted by Artw at 4:24 PM on May 19, 2010

Busy day for Google. Meet Google's copy of Amazon S3.
posted by Blazecock Pileon at 4:27 PM on May 19, 2010

Google Wave: Now open to the public
posted by Artw at 4:36 PM on May 19, 2010

Apparently Day 2 is indeed Android day.
posted by Artw at 4:37 PM on May 19, 2010

it wouldn't have anything to do with benefiting their consumers

I disagree. I know several consumers that is of great benefit too, me included. It would allow me, as a consumer, to view video in Firefox. It would allow, me, the consumer, to view video on a number of mobile devices. I get it though, those consumers don't count.
posted by juiceCake at 5:11 PM on May 19, 2010

I'm sure when the lawsuits pop up juries in the Eastern District of Texas will surely be swayed by "Google sed it wuz free of royaltees!"

Meanwhile back in here in reality I'm wondering whether Mozilla and Opera are joining this and exposing themselves to the mother of all patent lawsuits out of spite, out of stupidity or out of sheer ignorance.

If I could draw political cartoons I'd surely create one with ostriches labelled "Mozilla" and another one labelled "Opera" with their heads in the sand and stereotypical hunter running towards them with an elephant gun labelled MPEG-LA. Oh and a park ranger called "Google" with their back towards the whole affair.
posted by Talez at 5:14 PM on May 19, 2010

Talez, what do you think they should do? Kind of between a rock and a hard irrelevancy aren't they?
posted by BrotherCaine at 5:22 PM on May 19, 2010

On further thought, what I'd like to see happen is Apple purchasing full rights, patents, everything they need to own H.264, and releasing it for free in response.

Win-win.

/pipe dream

Does that even count as a pipe dream? Shouldn't pipe dreams be things merely as improbable as, "Maybe tomorrow I'll wake up and it'll be raining candy and adulation from people I respect and find attractive"?
posted by Copronymus at 5:42 PM on May 19, 2010

"a patent time-bomb waiting to happen."

On2 has been doing this since long before H.264 came out and they have a large patent portfolio of their own. In fact Youtube originally used one of their codecs before it used H.264.

Given this and the financial backing of Google, I'd say patent lawsuits could go in both directions.
posted by eye of newt at 8:00 PM on May 19, 2010

and the H.264 crowd has more to lose. What would Google lose? The right to give something away for free?
posted by eye of newt at 8:02 PM on May 19, 2010

OK, so hopefully my talk will be up soon. It should be on youtube.com/googledevelopers whenever they get around to uploading it (I'm the YouTube guy, who probably looks uncomfortable as I'm not used to this kind of public speaking :) ).

I can't address all the codec questions since I'm more on the HTML5 side. The On2 guys obviously believe they have a good codec, and Jim from On2 goes into more detail on the technical aspects and their own comparisons with h.264 in the video. One thing to keep in mind is that encoder settings matter a LOT with PSNR and other comparisons. The WebM stuff is all available on webmproject.org, so you can try it out yourself (well, assuming you know enough to make the pieces work). The tools link has info on how to patch ffmpeg to work, a DirectShow filter, etc. I'm sure we'll see additional comparisons by the public, now that they should have the tools to do so.

Browser support is looking good, Microsoft today announced that IE9 will support VP8 via DirectShow, so it will be possible to play these files in IE9, although you will need to install WebM support soon (which will be made easier in the near future). Thats 4 major browsers (Chrome, Firefox, Opera, and IE9) which should have WebM in HTML5. Links to the bleeding-edge versions with WebM are on the WebM site, I'm also going to update the YouTube HTML5 page with links.

YouTube started encoding new uploads of 720p or above in WebM a few weeks ago. We will be expanding this to all uploads soon. We're currently only producing 360p and 720p formats. This is to fill out the volume of videos, but we will add our other formats eventually. We get a days worth of video every minute, so we're ramping up our transcoding a little carefully. We will be continuing to produce h.264 transcodes as well.

Brightcove was also in the breakout session, and they'll be supporting it in their transcode pipeline and player as well, for all the sites that use them as a white-label provider.

We're working hard on our HTML5 player at YouTube, it's still got some rough edges but you should see improvements every week. Even the monetized video restriction won't last forever. You can find the webm videos by adding &webm=1 to a search result (such as a search for kittens). We'll also add that to the Advanced Search options (something that simply slipped my mind until yesterday), since adding it manually is ugly.

Obviously hardware support is an issue for any new codec, but we do have a (to me at least) quite impressive array of hardware partners.

[And hopefully not to disappoint anyone who expected otherwise, but I won't touch patent questions with a 10 foot pole, or lawyers will eat me. I'm just an engineer.]
posted by wildcrdj at 8:45 PM on May 19, 2010 [17 favorites]

Oh, and I totally forgot to mention in that, but Adobe will be supporting VP8/WebM in Flash as well. So while this gets tied in with HTML5 a lot (even by us), it's not limited to that, and that will help reduce the need to double-encode for people who don't need to reach absolutely everyone (obviously you'd need h.264 for older version of Flash, older browsers, some hardware platforms, etc).
posted by wildcrdj at 8:49 PM on May 19, 2010 [1 favorite]

One more thing I forgot -- in addition to new videos, we will be transcoding our existing catalog to WebM. But of course with 5 years of videos and the rate of new uploads, it will take a while to catch up. The eventual goal is to have everything available in WebM though.
posted by wildcrdj at 8:56 PM on May 19, 2010

I noticed after switching to html5 beta and randomly picking videos, many of them were still being served by flash player. I did find some that use html5. Are these using vp8? Or is that conversion process independent of webm?

Also: Is there a way to tell for sure which I codec I am using? I use Chrome 6.0.401.1dev, but I am not sure which codec is being used when I view a video found with &webm=1. I assume it'll transparently fall back to h.264 if my browser isn't supporting it yet.
posted by cj_ at 12:33 AM on May 20, 2010

and found it to be inferior to H.264 but better than Theora.

Of course printing out the frames on a dot matrix printer and making them into a flipbook would give superior performance to Theora.

Can you play MP3's on iPods and Macs? It's an inferior format to FLAC and AAC and yet it's supported.

Comparing MP3 with FLAC is absurd. AAC may be superior but there was huge range of existing content already available in MP3 format, that wouldn't be the case with a new format.
posted by atrazine at 1:29 AM on May 20, 2010

How is MP3 inferior to FLAC? One is designed to be lossless, the other to compress well. These are completely incompatible concepts.

I don't find Apple Lossless to be better than FLAC in any technical way, and I'm not really impressed with Apple's lossy format compared to MP3. LAME has done some amazing things with MP3 with their psychoacoustic modeling algorithms. I can't even tell a -V3 encoded MP3 from lossless, much less a -V0. I'm not sure what AAC brings to the table here.

I guess its status as proprietary is a strike against it, but Apple's codecs aren't open standards either. Has MP3's proprietary status ever really been a problem? MP3 decoding support is ubiquitous.
posted by cj_ at 1:42 AM on May 20, 2010

If Ogg Vorbis hasn't fixed that problem with sibilants and fricatives turning into mush even at high bitrates, I'm not going to like this one god damn bit.
posted by seanmpuckett at 5:17 AM on May 20, 2010

Even the monetized video restriction won't last forever.

I wonder how many people even know that the music videos on YouTube won't play on an iPad / iPhone. Anything that is from Vevo is not supported.
posted by smackfu at 6:13 AM on May 20, 2010

froyo officially announced

Android 2.2 'Froyo' beta hands-on: Flash 10.1, WiFi hotspots, and some killer benchmark scores

Flash 10.1 for Android beta unveiled: Hulu a no-show, Froyo now a minimum requirement

Google adding over-the-air app installation and iTunes streaming to Android

Possibly worth an FPP in itself.
posted by Artw at 11:00 AM on May 20, 2010

Heh. I was able to guess the exact selective quote Gruber would use. Well, that's not exactly hard really (it'll probbaly be popping up in one of the Apple threads any second now as well).
posted by Artw at 11:11 AM on May 20, 2010

The first in-depth technical analysis of VP8

Overall, VP8 appears to be significantly weaker than H.264 compression-wise. The primary weaknesses mentioned above are the lack of proper adaptive quantization, lack of B-frames, lack of an 8×8 transform, and non-adaptive loop filter. With this in mind, I expect VP8 to be more comparable to VC-1 or H.264 Baseline Profile than with H.264. Of course, this is still significantly better than Theora, and in my tests it beats Dirac quite handily as well.

Supposedly Google is open to improving the bitstream format — but this seems to conflict with the fact that they got so many different companies to announce VP8 support. The more software that supports a file format, the harder it is to change said format, so I’m dubious of any claim that we will be able to spend the next 6-12 months revising VP8. In short, it seems to have been released too early: it would have been better off to have an initial period during which revisions could be submitted and then a big announcement later when it’s completed.

Update: it seems that Google is not open to changing the spec: it is apparently “final”, complete with all its flaws.

In terms of decoding speed I’m not quite sure; the current implementation appears to be about 16% slower than ffmpeg’s H.264 decoder (and thus probably about 25-35% slower than state-of-the-art decoders like CoreAVC). Of course, this doesn’t necessarily say too much about what a fully optimized implementation will reach, but the current one seems to be reasonably well-optimized and has SIMD assembly code for almost all major DSP functions, so I doubt it will get that much faster.

I would expect, with equally optimized implementations, VP8 and H.264 to be relatively comparable in terms of decoding speed. This, of course, is not really a plus for VP8: H.264 has a great deal of hardware support, while VP8 largely has to rely on software decoders, so being “just as fast” is in many ways not good enough. By comparison, Theora decodes almost 35% faster than H.264 using ffmpeg’s decoder.

Finally, the problem of patents appears to be rearing its ugly head again. VP8 is simply way too similar to H.264: a pithy, if slightly inaccurate, description of VP8 would be “H.264 Baseline Profile with a better entropy coder”. Though I am not a lawyer, I simply cannot believe that they will be able to get away with this, especially in today’s overly litigious day and age. Even VC-1 differed more from H.264 than VP8 does, and even VC-1 didn’t manage to escape the clutches of software patents. Until we get some hard evidence that VP8 is safe, I would be extremely cautious. Since Google is not indemnifying users of VP8 from patent lawsuits, this is even more of a potential problem.

But if luck is on Google’s side and VP8 does pass through the patent gauntlet unscathed, it will undoubtedly be a major upgrade as compared to Theora.
posted by Blazecock Pileon at 3:17 PM on May 20, 2010

You mean the link in the FPP BP?
posted by Talez at 3:52 PM on May 20, 2010

Whoops. Missed it.
posted by Blazecock Pileon at 4:25 PM on May 20, 2010

Also: Is there a way to tell for sure which I codec I am using? I use Chrome 6.0.401.1dev, but I am not sure which codec is being used when I view a video found with &webm=1. I assume it'll transparently fall back to h.264 if my browser isn't supporting it yet.

We did add some badging on YouTube. The HTML5 player has a "HTML5" logo next to the format selector. If it's using WebM, there's also a "WEBM" logo next to the HTML5 logo. If you see just the HTML5 logo, it's h.264.

I don't know if the Chrome dev channel has WebM yet, the Chromium nightlies do for sure.
posted by wildcrdj at 5:58 PM on May 20, 2010

Nero Files Antitrust Case Against MPEG-LA
posted by Artw at 2:41 PM on May 24, 2010 [1 favorite]

How is MP3 inferior to FLAC? One is designed to be lossless, the other to compress well. These are completely incompatible concepts.

Lossless != incompressible.
posted by blue_beetle at 2:48 PM on May 25, 2010

In case anyone returns to this thread, here's the talk. Also, the WebM website blog has some new technical commentary, and the site as a whole has more resources.
posted by wildcrdj at 1:12 PM on June 9, 2010