Ogg Vorbis / Theora Language Removed From HTML5 Spec

Rudd-O writes "It's official. Ogg technology has been removed from the HTML5 spec, after Ian caved in the face of pressure from Apple and Nokia. Unless massive pressure is exerted on the HTML5 spec editing process, the Web authoring world will continue to endure our modern proprietary Tower of Babel. Note that HTML5 in no way required Ogg (as denoted by the word 'should' instead of 'must' in the earlier draft). Adding this to the fact that there are widely available patent-free implementations of Ogg technology, there is really no excuse for Apple and Nokia to say that they couldn't in good faith implement HTML5 as previously formulated."

...now that I read the changes...

[drakaan]

I see that what I just suggested is exactly the change they made. I'm fine with that...off to tag the front-page article with "badsummary"

Wierd.

[ak3ldama]

From the page [html5.org]:

It would be helpful for interoperability if all browsers could support the same codecs. However, there are no known codecs that satisfy all the current players: we need a codec that is known to not require per-unit or per-distributor licensing, that is compatible with the open source development model, that is of sufficient quality as to be usable, and that is not an additional submarine patent risk for large companies. This is an ongoing issue and this section will be updated once more information is available.

What part of initially suggesting Ogg Vorbis doesn't fit with the new quote? It just seems wierd. Like they could say what they mean, but not explicitly suggest Ogg.

Re:Who the hell is Ian?

[Penfold1234]

I assume it's referring to Ian Hickson, who's a member of the HTML5 working group.
If not, then I have no idea...

The only reason that I am even in a position to guess this is because I happened to go to University with him, so I agree the summary could use some work.

Patent FUD at fault.

[Anonymous Coward]

The actual removal can be found here [html5.org].

"we need a codec that is known to not require per-unit or per-distributor licensing, that is compatible with the open source development model, that is of sufficient quality as to be usable, and that is not an additional submarine patent risk for large companies."

The sad thing is that Ogg/Theora is strong on all these points, and it's probably the only somewhat modern codec set that even comes close. Theora might not be state of the art, but it is orders of magnitude better 1980s tech that someone might propose as an alternative (and Vorbis clearly is a state of the art design).

Meanwhile the MPEG LA licensed codecs that Apple and Nokia are advocating have already landed several *licensees* in court for patent litigation, with two major cases ongoing. In particular the MPEG LA license agreement is quite specific that the license does not provide all the patents needed to implement the covered codecs. Some of the lawsuits have even been from members of the pool (such as Lucent), so paying up provides you with little protection from attack from the pool members, no zero protection from patent attacks by third parties.

Theora and Vorbis were designed to be free of serious patent problems. That doesn't mean that they are completely immune, *nothing can be* in our current patent climate. However, they should do better than their proprietary competitors... and the track record shows that.

The actual mail on the HTML-wg mailing list...

[Anonymous Coward]

Just to point out what it currently happening, here is the mail from Ian Hickson from this morning:

"I've temporarily removed the requirements on video codecs from the HTML5
spec, since the current text isn't helping us come to a useful
interoperable conclusion. When a codec is found that is mutually
acceptable to all major parties I will update the spec to require that
instead and then reply to all the pending feedback on video codecs.

    http://www.whatwg.org/issues/#graphics-video-codec [whatwg.org]
"

The title of the news is a bit misleading :) In other words "temporarily removed until a consensus has been found".

You're a bit off.

[Anonymous Coward]

Theora is a bit computationally cheaper to decode than H.264, and of much higher quality per bit than older generation codecs.

W3C was quite clear in their diff that pay-per-use codecs like H.264 were utterly unacceptable to them. This isn't a choice between Theora and H.264, it's a choice between Theora and H.261 ... which needs on the order of 10x the bandwidth to still have worse quality than Theora, enough of a difference that H.261 is not really suitable for web streaming. So it's really a choice between Theora and nothing at all, a state of affairs that screws the public but should leave the codec licensing folks happy since they are already making great money off the fragmented status quo.

Theora is not as good as H.264, but it's not that far behind, and it's much better than anything else no-cost. For a baseline codec it doesn't matter that chosen codec isn't the best quality available, it matters that it isn't terrible and it matters that it can be universally implemented. Today Theora is pretty much the only option that meets those two simple criteria.

And, of course, Vorbis is a state of the art codec which stands up well even next to the best AAC-HE codecs, even at low bitrates.

Re:Doesn't make sense...

[gnasher719]

FYI yes its patented but its under a 'free for all' licence. Anyone can use it for any purpose for free (both beer and freedom).

Who knows if there are any other patents lurking? We know from the MP3 history: Yes, everyone knew there were tons of patents, but you could buy a license for all the patents for quiet a reasonable fee, which for example Microsoft did. Then suddenly someone comes up with a claim for a patent that is not covered by that pool and sues Microsoft for billions.

Same thing could happen with Ogg as well. Make it part of the HTML5 standard, convince Apple, Nokia and Microsoft to use it, and four years from now when every browser in the world supports it, someone comes out with a patent claim and sues everyone. It doesn't happen right now because trying to enforce a patent on Ogg won't make you any money.

FUD FUD FUD

[a known emus]

This is a point by point reply to your FUD.

• Theora is almost an order of magnitude better performing that H.261 and this is a critical difference for web video.
• No one actually knows what the patent status of any software is! ... In fact, several paid up licensees of mpeg codecs have been sued for patent infringement over these codecs *and lost*, so it's hard to argue that those codecs are better off.
• It's true that Theora isn't very widely adopted, but it has been shipped by Linux distributions for years, so there has been plenty of opportunities for people to sue over patents. Theora is used by Wikipedia, one of the most viewed websites in the world. Of course, Vorbis is orders of magnitude better on this point.
• What are you Nokia? An expensive and heavily patented codec like H.264 is not "open" in any meaningful sense. It's true that Ogg/Theora+Vorbis is not yet amazingly popular, but that is part of the point of standards. There is a chicken and egg, and first-mover takes all problem for file formats and standards help fix that problem.
• Why does HTML have an image tag? What would the world be like if images on the web required various incompatible proprietary plugins? Why should video and audio be any different from still media?

Re:An alternative...

[cching]

What are you seeing as the big reason you'd have to supply media in multiple formats?

I have to provide multiple formats because it's hard for users to find the right codecs to be able to consume my media. *You* don't think it is a problem, but I'm telling you that in reality, it *is*. For instance, some media players don't support MPEG-4 but they do support .WMV. And it's just the opposite for other players. I don't know, I guess I thought everyone understood this. Audio is slightly better, but I still run into the same problems. There's a reason that sites like amazon.com provide two different audio formats so that people can preview the music. Welcome to 1998, it really hasn't gotten much better since then. I do believe that standards would make this better.

Re:mod parent up.

[Bogtha]

What about standards for video and audio rendering? Why don't these exist?

They do [wikipedia.org]. However having standard codecs and attempting to tie together formats that are orthogonal in nature are two separate matters. The <video> element type works regardless of the video codec in use, so the HTML specification attaching itself to one particular codec is unnecessary.

Re:FUD FUD FUD

[Just Some Guy]

And countering yours:

• From it's inventor [mit.edu]:

  Unlike Vorbis and Speex, legitimate best-in-class codecs, Theora's coding quality is obviously poor relative to contemporary competition. This poor performance stems both from implementation and design deficiencies. As a seperate problem, Theora is also poorly integrated with Ogg due to incomplete multiplexing software and documentation on the Ogg side. Without guidance from Xiph.Org, outside development and implementation of Theora-in-Ogg has been chaotic and of low quality.

• It's safe to say that MPEG4 and it's codecs have been more thoroughly researched than Theora. Remember the FOSS mantra: "many eyes make all bugs shallow"? That applies to lots of things, such as many video producers' legal teams checking this stuff out.
• I absolutely, positively promise you that Youtube serves more video than Wikipedia, and they don't stream Theora.
• You're imagining that Theora is equivalent to H.264, etc. It's not. There's no first-mover advantage to it because it's already been overtaken by, well, pretty much everything.
• There's no standard web image format. By convention, most people use GIF and JPG (with a few PNGs sprinkled about for good measure), but that's just the way it happened to work out. I'm not sure why people have this wrong impression, but it's simply not true. Don't believe me? Read the spec [w3.org] yourself. If that isn't clear enough, W3 explicitly states [w3.org] that

  The HTML specification does not prescribe or limit which graphics format you can use.

I'm a huge FOSS buff, but that doesn't mean I have to blindly love everything pushed out the door as "freedom friendly". I don't have anything against Theora except that it's just not very competitive. I wouldn't want to see it as the official video file format any more than I'd want to see ASCII text as the official document file format; both have clear limitations when compared to their competitors.

The W3 made the right choice. As much as I like the idea of Theora, I'm glad we don't have to be saddled with the reality of it.

Re:Figures

[Bogtha]

You correctly note that current the HTML 4.01 recommendation doesn't require JPEG, PNG, etc., but you fail to note that it does specifically mention three image formats, and they are "GIF, JPEG, and PNG".

Yes, it mentions them, it doesn't recommend them. Look at what it says:

src = uri [CT] This attribute specifies the location of the image resource. Examples of widely recognized image formats include GIF, JPEG, and PNG.

It mentions them as examples to illustrate how the <img> element type is used, not in order to promote them and certainly not to "specify them as baseline standards" as Ignorant Aardvark was claiming.

And how in hell did I "fail to note" that it mentions them? I explicitly said it mentions them.

Of course, if you did mention that, it would be a lot harder to use the current recommendations treatment of images to argue that removing the mention of Ogg formats from the HTML5 draft is consistent with the way prior HTML standards have treated images.

That's simply not true. There is a world of difference between mentioning popular formats as examples and saying that vendors should implement them.

Re:Ummmm.....

[evilviper]

It's comparable to H.261 in performance

That's only according to some guy from Nokia, who clearly has a massive bias.

No one actually knows what the patent status is

On2 had sold VP3 licenses for years, as well as for newer versions (VP4/5/6/7) based on many of the same methods as VP3. Those codecs have long been licensed, and widely used by very large companies like AOL (Nullsoft TV, AIM Video), Macromedia, Adobe (Flash v7), BBC (QuickLink field broadcasts), eBay (Skype Video), and no doubt many many more. The fact that no patent trolls have come out of the woodwork yet is pretty damn strong evidence that Theora is in the clear.

No one even uses Theora for anything

That's pretty much how all video codecs start out... Theora is still in beta, yet there is quite a bit of content from sites such as http://v2v.cc/ [v2v.cc]

Why do we need video requirements for text markup?

For the same reason we need image requirements for text markup. In fact HTML already has a specified video format: raw MJPEG, it just happens to suck.

I long wished MPEG-1 had been specified for web video (to supersede MJPEG) when it's patents had first expired, but it never happened, no doubt because some many companies have vested interests in getting those patent license fees.

Dolby Labs does the same thing whenever a video standard is being defined... they throw a good amount of money in bribes around, and make sure the standards (in all those countries that have software patents) include only Dolby, despite MP2/Musicam being as good, and trivially easy to include as an alternative. So while our European friends can put free MP2 audio on their DVDs, and in their DTV broadcasts, in the US we are absolutely required to have a Dolby Digital/AC3 audio track.

OT: Possible Linux Firefox/Flash Crash Workaround

[beej]

Firefox crashes on me every time I view a YouTube video thanks to the wonderfully crappy proprietary, closed source Flash plugin for Firefox on GNU/Linux.

Hey--I used to have a problem where Firefox (tried many versions) under Linux (Slackware 12) would hang consistently after viewing a flash movie (tried many plugin versions) then starting to view a second one. I suspected it had something to do with the audio, I found after tracing with gdb, and seeing how non-audio flash pieces didn't crash it.

Turns out that if I start up xmms or audacious on some mp3, then hit pause (has to be pause--not stop) or just let it play, then I can view flash movies to my heart's content without a crash.

It's a kludge, but maybe with a little luck it works for you.

Re:mod parent up.

[Sancho]

100% correct.

I liked the idea of OGG being recommended for HTML5, but realistically, there are a lot of problems.

As a container, OGG is pretty heavyweight. It's not going to be good for mobile devices.
All off the Vorbis and Theora decoders I've seen have been extremely resource-intensive. This may well be because more attention is devoted to other codecs like XVid, and so they are more highly optimized. Nevertheless, again, mobile devices will suffer.
Quality-wise, Vorbis is pretty nice. Theora, however, is a generation behind, and rapidly losing ground. HTML5 isn't expected to be ratified for over a year. In that time, Theora's generation of codecs will be even older and less efficient to the then-current codecs. For a field as rapidly evolving as streaming video, it doesn't make all that much sense to include it. It would be like suggesting that Indeo be implemented for HTML4.

The biggest benefit to recommending OGG in HTML5 is that it would get a free format out there, but at the cost of efficiency. While bandwidth continues to grow, and computers get faster and faster, waste is still a concern, and mobile devices are becoming more popular (you have to treat these as if they were 10 year old computers with equivalent bandwidth!) OGG misses the mark in most categories--too big and bulky for mobiles, too old for new computers. It's the worst of both worlds.

Re:Yeah, that's FUD

[xiphmont]

"That didn't come across so well in the presentation. That is the qualification wasn't apparent - it sounded like you were saying it was bad, period."

That document was an internal identification of weak areas. It was intended to be 100% critical. We don't need to keep telling ourselves within Xiph and RedHat etc what's good about it.

"But the legal teams of every major corporation that wants to use it [MPEG]? I imagine Sony's given it a look or two along the way. Just as an example."

Sony is a bad example as they've attempted to undermine MPEG with a number of spectacularly ineffective market fragmentation tactics. However, the major supporters of MPEG tend to be MPEG themselves. Those who use MPEG want to avoid Microsoft. Real is widely perceived to be dying. We're considered a risk, mainly because MPEG says so. There are no other options left.

"Honestly, I wish you well - we're on the same side!"

np. I just don't deal in weasel-words is all...

"but I respectfully disagree that Theora's ready to be standardized right now today."

Not true. It's been mature/ready for a very long time. VP3 is actually slightly older than Vorbis. It fills its niche perfectly ; its theoretical performance, which we're closing in on, is very good compared to how much more 'modern' (but much heavier) codecs perform today, and it does it at a fraction of the complexity. It's perfect for lightweight implementations and ideal for the tag.

Re:Ogg/Theora only poor against impossible options

[evilviper]

The only real candidate that I'm aware of is H.261, which takes roughly 10x the bandwidth of Theora to deliver anything close to the same quality at typical web streaming sizes.

First off, MPEG-1 is no longer patent restricted, and is newer and better than h.261.

Second, 10X is clearly a made-up number. Through the past 20+ years of lossy video compression, there hasn't been an order of magnitude improvement in compression at all. And even if there had been such an improvement, Theora certainly wouldn't be the codec in a position to do it, as it's pretty poor quality. If you're really seeing that huge of a difference, you're doing something HORRIBLY wrong.

I have tried h.261 even though support for it is pretty flaky, and I use MPEG-1 EXTENSIVELY today (on SVCDs and DVDs, in lieu of MPEG-2). I've got a video encoding to MPEG-1 right now... I would put libavcodec's MPEG-1 up against Theora any day. If nothing else, the quality is quite close, and MPEG-1 requires a tiny fraction of the CPU power to encoder or decode.