Slashdot Mirror
Oh, What a Lovely Standards War
ChiefMonkeyGrinder writes "You know something big must be afoot when people start to get worked up over video compression standards. Basically, the issue is whether the current de facto standard, H.264, will continue to dominate this field, and if not, what might take over."
Related, reader eihab writes "Nuanti, a company that develops Web browsing technologies, has produced a high-performance Ogg Theora decoder for Microsoft's Silverlight browser plugin. Nuanti's Highgate Media Suite will enable support for standards-based HTML5 video streaming with Theora in browsers that have Silverlight. It works entirely without requiring the users to install any additional software."
It works entirely without requiring the users to install any additional software."
Except, of course, a browser that has Silverlight. :-|
Currently hooked on AMP
...Silverlight
it's just as bad as flash only from an even scummier company.
The only video codec that every browser can use at the moment is Ogg Theora. Unlike H.264, there are no costs involved beyond implementing support for it in your browser and there are no licencing issues that prevent distribution. Firefox, Opera, and Chrome currently support Ogg Theora. It's a shame that Safari and IE won't support it by default in the near to medium term.
It will be interesting to see what Google does once they own On2 Technologies. They may choose to open source the VP8 codec so every browser can use it and make it the default codec for YouTube, possibly as VP8 in Ogg as Ogg is a pretty good container format.
For now, the Video for Everyone code hack is the solution. Works on Firefox, Opera, and Chrome natively with Ogg Theora, and Safari natively with H.264, and Internet Explorer with Flash (loading the H.264 content).
Naturally the best solution would be that everyone implements Ogg Theora as a standard fall-back solution, and use their "better/proprietary" solution when available.
I wish there was a way to mod the original press release as +5, Epic Troll, because that's what it is with respect to Slashdot - it's going to be way more entertaining than the usual (and already somewhat tiresome) Google vs "do no evil" stories. But Microsoft's Silverlight used to enable support for Theora in pretty much all Windows browsers (and specifically IE of all things), while both Google and Apple stand by H.264 - oh my!
Hold on a second, I've got to fetch the popcorn...
"It works entirely without requiring the users to install any additional software."
Other than Silverlight. Gee, that solves the problem.
Ogg Theora won't become relevant until there are hardware decode chips available. Why would I install Silverlight to play Ogg when I can use HTML5 and H.264 instead? Because someone might charge to develop with the codec after 2015?
I don't care because the H.264 standard is open even though it's not free.
By virtue of the de facto status, it seems like anything that the majority of people use will never be superceded by anything that barely matches or only slightly improves on the de facto standard. From what I've read Theora is quite bare-bones compared to H.264 and hasn't been designed with hardware decoding in mind.
On a more technical side, I found this bit in TFA interesting:
We'll be releasing a high-performance decoder for Theora video/Ogg Vorbis audio streams that plugs into the Silverlight 3 streaming media abstraction ...
I know little about Silverlight, only the most general look and feel, and capabilities. Does this mean that it actually has extensible codec framework, that can be extended from managed code (since any SL code has to be managed, so that it can be properly sandboxed - same as Java applets which cannot e.g. use JNI)?
If so, the next logical question is - can the same thing be done with Flash, architecturally?
As a side note, this also means that Silverlight CLR JIT produces code that's fast (not just "fast enough", but actually "high-performance", at least if the claims are true) for a video codec, which is quite impressive. I'm not sure you could reach the same levels with ActionScript, due to its inherently dynamic nature, even with Adobe's JIT. But perhaps I'm underestimating the ability of modern JS JIT compilers to do static type inference, and consequent optimization based on that type information?
Either way, pragmatically, this means that any browser running on Windows will be able to play Theora after installing Silverlight - which, by the way, pops up in "recommended updates" list in Windows Update as soon as you install Windows. While Silverlight plugin is only officially supported on Windows in IE and Firefox, IIRC, I haven't had any problems using it in Opera regularly, and I've seen it work in Chrome, so it does seem to be mostly browser-agnostic.
It would be very ironic if Chrome running under proprietary Windows and OS X could play Theora, while Chrome on Linux would only support H.264.
But somehow, I don't think that will matter. Ultimately, Google is the 800-pound gorilla here because of YouTube, and most likely whichever they will go with (and they have already said they want H.264) will become the de facto standard. Apple could probably steal the day, but they stand by H.264 as well...
If they did, everybody could just use that (since it's already on 98% of computers out there) and put a stop to these stupid standards wars.
.jpg/.png editing software.
They probably wouldn't lose much revenue, if at all... I mean, they've always been giving away the Flash plugin for free. They make all their money from selling content-creating software (Flash CS3) right? That wouldn't change if they open-sourced Flash player. Similar to how Photoshop completely dominates the industry even though anyone is free to make
MKV files don't work on bloody anything reliably except VLC, even though they're theoretically an h264 variant.
Matroska (.mkv) is not a "H.264 variant". It's not a codec at all! It's a container format, which usually contains an H.264 video stream these days, but this has varied historically, and is not in any way standardized.
HTML5 is a markup standard. Where it pertains to video is in the standardization of video-related markup, i.e. the "video" tag, not video formats. W3C has nothing to teach MPEG about video formats. W3C also has nothing to teach MPEG or ISO about standardization, because the Web is a mess of proprietary IE and Flash while MPEG has enabled 20 years of consumer digital video, including the DVD and Blu-Ray. Right now, both QuickTime Player and FlashPlayer play H.264, both iTunes and YouTube are H.264, both Flip and iPod camcorders are H.264, but I can't make one Web app for both IE and Firefox.
What we are talking about with Web video today is "will our H.264 video playback move from plug-ins (QuickTime Player and FlashPlayer) to native browser playback?" That is all. The format is not in question. The HTML4 Web has already been using the ISO standard format in iTunes, YouTube, and many others. There is no competing format. FLV is still used too much, but it has been deprecated since 2008, it has no HD sizes, it is proprietary to Adobe, the encoder costs $599, and it takes much more bandwidth than H.264. There are no Ogg camcorders, iPods, video editors. These tools and devices were all built for MPEG-4, which is a standardization of the QuickTime file format that was used previously. Google has already said that even if they had the compute time to transcode YouTube to Ogg, the Internet does not have the bandwidth for an Ogg YouTube, and almost nobody has a player.
Ten years ago, Linux users complained that they could not view the video on the Web because it was in QuickTime containers with Sorenson video and Qdesign audio and that was all proprietary, not standardized. Now, the video is all in ISO MPEG-4 containers, with ISO H.264 video and ISO AAC audio and is playable on Linux in FlashPlayer and WebKit browsers and other players, and the complaining continues. It is disheartening.
Nuanti has produced a high-performance Ogg Theora decoder for Microsoft's Silverlight
Hardware accelerated H.264 is in the 10.1 Flash Beta. Silverlight 4 will support Chrome. The "high performance" H.264 player will be everywhere and in everything in the next few weeks or months.
Yes.
For one thing, I dispute your assertion that Google is doing evil. It's a company, it's doing what's in its best interest. Still, I know of few companies who have contributed so much to open standards and yes, even open source software, to the technological community. But I digress...
For another, it boils down to one simple question. Do we want a de facto web standard to depend on a patent-encumbered standard? We've been there before. Remember the GIF kerfuffle? Remember the JPG morass? Remember how long it took Microsoft to get a browser out there that supported PNG, which is a better image rendering codec, and a standard that all browsers (or any other software developer, open source or otherwise) can implement? Wouldn't it be nice if, just for once, we could bypass all of the stupidity and just settle on something up front that's easy and that everyone can support?
Also, what the hell good does it do to write a web standard designed to get people out of the Flash embedding hell that we're in right now, only to put us into yet another hell of a patent consortium that may or may not charge exorbitant fees to develop software with its standard built in?
Let's not fool ourselves. Anyone who wants H.264 to become a web standard, whether it be codified or de facto, is basically saying, "I hope that Firefox dies a miserable death." Why? Because Firefox is open source, and as such, it can't build in a patent-encumbered codec like H.264. On the other hand, most other browser makers (Apple, Microsoft, Google, etc.) have a choice. There is nothing stopping them from implementing either or both. "Evil" Google has taken the middle road, and Chrome handles both. We have yet to hear from Microsoft. Apple has chosen to deliberately not support Ogg Theora, even though it would be trivially easy for them to do so. It has taken this position, I believe, because it knows that Firefox can't implement H.264. I honestly think they want to kill off Firefox so that there's more marketshare for Safari.
So yeah, I will gladly cast my lot with those who support Ogg Theora as THE video standard of the web, and I don't care who they are. If Microsoft wants to come on board, then hell yeah, +1 for finally doing something right and that will ultimately benefit all video producers and consumers. Google and Firefox already have Ogg Theora built in, so they've already earned their +1, even though Google was one of the objectors to Ogg Theora being codified in the HTML 5 standard. I've given up on Apple getting anything but -1 Troll for this issue.
Wow, got a flamebait in record time for that one.
No offense to the OGG crew and developers, but what you're not getting is that the battle is already lost. The future of web video isn't really in the browser. It's on low-powered appliances like XBoxes, iPhones, iPads, Playstations and the like. And that's now. People are already building libraries in h264 and divx because of this. It's an insurance policy against your media not becoming obsolete like VHS and DVD.
Divx just slides in because most devices will play it hardware assisted even though you need to install the codecs on a desktop.
Without hardware decoding on those low-powered devices, and the ability to play your media anywhere you damn well please with no software installs necessary and no transcoding required, you may as well not exist.
OGG's a fine set of codecs, but if I have to transcode out of it to play on anything but a desktop, basically, I have no use for it and neither does the consumer other than the idea behind it is a quite appealing one.
Now, the video is all in ISO MPEG-4 containers, with ISO H.264 video and ISO AAC audio and is playable on Linux in FlashPlayer and WebKit browsers and other players, and the complaining continues. It is disheartening.
The complaining continues because Linux users still cannot play video using FOSS solutions, due to licensing fees associated with implementation of H.264. Given the overall Linux philosophy, it's a perfectly valid complaint.
It seems to do mostly with the ability to use more audio codecs in MKV, and more subtitle formats. In theory MP4 is just as extensible there, but in practice few players understand IDs for anything but the "official" codecs - which do not include e.g. AC3.
MKV is a container. OGG is container. H.264 is a codec.
Basket vs Fruit.
For large sets, this will be our guide even unto death, for the LORD will work for each type of data it is applied to...
Theora can be decoded on the cpus of all the devices you listed, at the applicable screen resolutions, in real time. Heck, the arm optimized version of theora can decode HD at a significant multiple of real time on a CPU slower than the one in the 3gs.
All of this craze and expectation of hardware acceleration comes from H.264 being an utter pig. They overestimated how much faster cpus and memory would become by now, and we're only coping by using lesser profiles or adding hardware acceleration.
mpeg-LA seems to be letting broadcasts go free for the next couple of years. Note that is only for the actual broadcast. They can open a can of whoop ass on various licensing fees whenever they feel it gets entrenched.
They can, but you know they will not until 2017 (expires in December of 2016). You can plan around and to a date.
Meanwhile Theora is an unknown patent quantity that may or may not be challenged at any time. It's the schrodinger cat of codecs, so no-one even wants to hold the box much less look inside.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
It's on low-powered appliances like XBoxes, iPhones, iPads, Playstations and the like.
The PS3 and Xbox 360 are enormously powerful. The original Xbox does not do a good job of playing H.264. The PS2 does not do a good job of playing anything. What were you saying, again?
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Yeah, except with MPEG-LA charging website owners a per-video fee (ensuring most webmasters avoid it) and with both Firefox and Opera refusing to implement it, h.264 already lost the battle as well. It's not about user's devices, it's about websites and no website will pay MPEG-LA's extortion fees and exclude over a fourth of desktop users and a significant part of mobile ones in the process.
It's been Theora or nothing from the very beginning. You argue that it's nothing, then, and I'd be inclined to agree with you, but the idea of h.264 becoming a web standard was dead on arrival. Which is, I suspect, exactly what Microsoft and Adobe wanted from the beginning as the status quo is what benefits them the most.
No problem is insoluble in all conceivable circumstances.
I don't know about handheld but MKV files work pretty well on my Western Digital TV thing. Plays back h264/aac, h264/vorbis/vobsub, mpeg2/ac3/vobsub, all of those in MKV containers.
Other set-top devices apparently have support for mkv files too (don't have any other set top boxes to test it on, the WDTV HD works too well for me to try anything else).
What are you talking about? The only thing preventing firefox from incorporating H.264 is the fact they don't want to pay a license fee to do so. Open source has absolutely nothing to do with it. If that were the case Chrome wouldn't have it included. I know you figured you'd get some more mod points by claiming firefox is somehow defending open source with their decision, but that's not even remotely accurate.
The thing about submarine patents is you don't know which technology is actually in the crosshairs until they pop to the surface. Anybody who has been sued by patent trolls will tell you that independent invention is not a defense. Neither is the excuse that you couldn't see the patent, or that the patent only tenuously describes what has been implemented if you squint just right.
The problem is there is exactly zipola when it comes to hardware acceleration for Theora, and you can't just "use moar power!" when we are talking cell phones, netbooks, and mini-tablets. I have gotten nothing but grief for pointing this out, but if Theora is to have a chance it needs hardware acceleration by the big three-Intel, AMD, and Nvidia, and it needs it yesterday.
My GPU cost a grand total of $35 and came with H.26x, WMV 7-9, MPG 2 and 4, all accelerated out of the box. Even the cheapest onboard GPUs nowadays usually will give you H.26x, WMV, and MP4 acceleration out of the box. And building PCs I can tell you it does make a difference, even on powerful machines the playback is smoother and allows for more multitasking without video stutter.
So I would be suggesting to the FOSS community if they want Theora to get anywhere they better look at the specs AMD and Intel have released and start on hardware Theora decoding ASAP. If they get it going for those two Nvidia won't allow themselves to be left out and will get Theora acceleration if for no other reason to have a "me too!" bullet point on the GPUs. Then you will have hardware acceleration covered for most desktops and netbooks, which in turn will hopefully make the cell phone and other small Internet devices stand up and start to take notice.
But just saying Theora runs okay without hardware acceleration on your desktop won't cut it, when so much of the Internet is moving away from simply sitting at a desk all day. Both AMD and Intel have released specs on their GPUs, and isn't that what the developers have always asked for? Give us the specs and we'll take care of the rest? Well here is your chance, we need Theora acceleration if H.264 isn't to become the dominant format.
ACs don't waste your time replying, your posts are never seen by me.
The price of admission is sending people to the four times a year MPEG meetings. The chips are the patentable intellectually property. The game is to get your IP into the standard by any means possible. When you are in the standard then you get profit participation in the MPEG-LA revenue stream.
When I was involved, the Japanese had a notorious reputation for sending lots of people and stacking the meetings. They would use procedural methods to extend the meetings into late night and then after others left they would use their numbers to force through their proposals.
Of course other players had other ways of stacking the deck. Remember that big corporations can afford to employ people full time to chair committees and that gives the extra clout (MicroSoft, apple, Sun, Philips,...).
This all means that smaller independent groups, like the one I worked for, had a very difficult time making any headway. No matter how good the technology, political considerations had a lot more impact.
The trick is that while MPEG is an open international body that supports "open standards", MPEG-LA is a foul black pit full of zombies, orcs and lawyers. In fact, the orcs and zombies are at the bottom of the heap, because the lawyer are the bad asses who run the show.
How are licenses fees set? Nobody knows. How are revenues divided? Nobody knows. How much is spent on MPEG-LA costs? Nobody knows. How do they decided to engage in legal action and who do target? Nobody knows.
It is a completely independent body with no oversight by any of the international standards bodies, or any government for that matter. It is only constrained by the software copyright rules in an individual jurisdiction.
It is a closed black box that can charge as much as it wants, and because it is an "international standard", it is almost impossible to compete with it based on cost or quality, and and you can't go after it using the legal system. (This one reason is why Ogg Theodora is not looked at as a meaningful option by the big players; it is not a standard, so it gives big companies headaches. Who is responsible if there is any trouble? What happens if a key person is hit by a bus? Having access to the source does not fully address all these legal issues.)
The reason that this such a bit deal is that large amounts of money are involved. I Googled around and I couldn't get a clue about total amounts, which is suspicious in itself. Remember, from the corporate viewpoint this is "free money", because the initial investment is small; a lab with some computers, some PHDs, a travel buget and some lawyers and the cost of their shark tanks. Very high rate of return over a long period of time.
And a shout out to all you libertarian morons out there: THIS IS A TAX!!! It is a tax collected by corrupt self serving insiders who have subverted the legal system. It restrains trade and stifles innovation. It is not subject to competition. Those who are taxed have no say in the matter. It is arbitrary, and you cannot escape it by taking your business elsewhere. It is all the things you claim to hate about government. How come you this behavior is good when done by business for greed and bad when done by governments, which are more accountable to the people?
Why is Snark Required?
Seriously... Vorbis has not even taken over MP3, despite it being far superior.
And you expect Theora to beat H.246??
The fact is, that apart from us few experts, nobody cares what format it is, as long as it works, and has the best quality for its size.
Look at what movies are used on BitTorrent nowadays. It’s mostly H.264, since the quality is simply superior. And XviD, since that’s what most pre-bluray standalone players can play.
Even though I’m a supporter of open formats, I support H.246 right now. Because there are two groups of sources I have:
1. Commercial video streams (YouTube, Daily Show, South Park, etc), who can handle the legal rights, and usually have a license anyway to distribute physical media etc.
2. P2P-shared movies, that don’t care for laws anyway.
(Bonus question: Guess how I would release my work? ^^)
But: Offer me something that has all features of H.246, plus only one single tiny superior property, and I’ll be the strongest supporter of that format, that you will be able to find.
Until then, it’s no war. Because one side has no teeth at all. (Sadly.)
Any sufficiently advanced intelligence is indistinguishable from stupidity.
Installs in Silverlight but doesn't require additional software?
Huh? That's full-on doublespeak.
No, that's merely assuming Microsoft will start bundling Silverlight with all new versions of Windows/IE sometime in the future. And given their history, particularly that of the .NET framework itself, that's a very reasonable assumption.
MSFT will never bundle Silverlight with IE or Windows. They've spent the last decade being sued in every anti-trust court in the world because they bundled IE with Windows. They won't make that mistake again.
Do we have to play words? We both know that, at the very least, a considerable proportion (I dare say, a majority) of Linux users prefer FOSS over non-FOSS, and at the very least, open standards unencumbered by patents (and associated fees) to closed ones. The fact that many of them still use proprietary software (and hardware with such) - NVidia drivers, Android etc - does not change that. It just means that sometimes, pragmatism outweighs purism. It's not black & white, after all.
It doesn't mean that they like that state of affairs, however. Back when GIF was patented, I haven't heard of anyone disabling that code in their browsers - but there was, nonetheless, a big campaign in support of a switch to PNG.
h.264 video outnumbers Theora video on the web by many orders of magnitude. Perhaps you missed the memo, but YouTube, Apple and Hulu all use h.264 extensively. Asserting that h.264 has somehow lost is delusional.
As it stands, h.264 is the dominant web format for new video, only possibly outnumbered by legacy videos (which are very much *not* encoded with Theora).
Claiming that 1/4 of the desktops on the web can't view h.264 is rather amusing given that the vast majority of Firefox installs play h.264 just fine, as they almost universally have the Flash plug-in.
MP3 and MPEG-2 users have both been nailed by surprise ligation for unknown patents which were not covered in the MPEG-LA licenses. The MPEG-LA license pack explicitly disclaims all warranties and rejects all representations that the license contains all the patents necessary to implement the formats, much less ones which might not be strictly necessary but that your implement ion might practice for good performance.
So, if you use H.264 you have to pay the license fees, plus take the risk that someone will sue you for practice some patent not in your coverage. With Theora you only have the latter risk.
Whatever argument you can make about increased confidence in H.264 due to wider deployment can be countered by the fact that its a much newer format during the design of which NO effort was taken to avoid patented technology (the MPEG and ITU processes both forbid IPR discussions during the main standardization process for anti-trust reasons). Whereas Theora is a more conservative design built from the ground up (by On2) to be free of third party IPR.
How can you possibly know what is contained in the ACTA treaty? While the actual contents remain secret (and they're likely to remain so until passed into law in the signatory countries) we have nothing to go on but leaked documents. Documents leaked already suggest the treaty aims to go well beyond simply protecting copyrights and trademarks. But of course all the countries involved in the negotiations already have laws to protect copyrights and trademarks so if the treaty didn't go beyond that there would be no reason for it to exist.
1) The main point really is that you can now relatively easily deploy Web video in Theora without sacrificing much potential user base. (Cortado can fill in some gaps in native browser support already, but Java applet support is dwindling.)
1a) It might not yet be default(?), but MS is actively pushing Silverlight for Windows users, so the installed base is already fairly large and growing.
1b) Apple I hear has some at least semi-official Moonlight-based support, but this I know less of. Comments?
1c) Though not the best in quality per bit, you can make the quality of any codec better with more bits. Bits are only going to get cheaper. H.264 can potentially get much more expensive.
2) No, H.264 won't die a gruesome death now.
2a) Yes yes, we all know it's better technically, it doesn't matter, it still can't be a baseline Web codec.
2b) Yes, some players, especially those with vested interest in the MPEG-LA racket and excluding smaller competitors, will almost certainly use H.264 on the web for a long time to come.
2c) Isn't it nice though that a widely deployable option exists that probably has already played a hand in how much money the MPEG-LA can squeeze from you if you _do_ decide to go with H.264 anyway?
3) Using H.264 for everything won't be as unified as you think.
3a) Much of the material on the web incidentally doesn't use the very advanced features of H.264, because many decoders are limited in what profile or subset of H.264 they support (thus also reducing the quality advantage to Theora, but I make no claim of its elimination)
3b) Some material (like pirated stuff that doesn't care for copyrights or patents alike) will use all the bells and whistles, but then you may well still be stuck with having to transcode for different devices even if everything does "H.264".
3c) Such conversions can be relatively well automated when needed while keeping the original not to incur generation loss; I don't really see some need for transcoding persisting as a huge deal, except of course to the extent that anything you do with a patented format might be illegal depending on jurisdiction and circumstance.
4) Yep, no "hardware" (DSP) decoders for Theora abound.
4a) Mobile devices have enough oomph to decode it anyway in relevant resolutions (Theora is lighter than H.264, too)
4b) Yes, battery life will probably suffer somewhat, doesn't make it useless.
4c) Some DSP work has already been done on Theora decoding as already previously commented, though even when ready, deploying it would probably require user intervention and sufficient access unless shipped by the OS itself. ("Install this to improve your battery life with this site.")
Hope this summary will clarify things somewhat.
Nah... java applets, trust me, it will WORK! This time...
Reminds me of a comment on a dutch tech site, remarked how much smarter a dutch tv station was, for choosing silverlight over flash, because it was more widely supported, except that particular function just happens to only be available for windows.
Silverlight may or may not be good, but after ActiveX and COM and such, why do people keep building their business model on an MS product? You know that sooner or later they will pull a move that screws you.
It would be like putting a bet on Apple announcing a sensible, non-sexy, non-drool inducing, cheap and essential item. Or IBM doing anything interesting in the consumer market. I don't know about leopards, but I do know companies never change their spots.
MMO Quests are like orgasms:
You may solo them, I prefer them in a group.
You question how Theora would be accelerated on Windows proprietary drivers.
We call them Shaders, and this can be done in regular old GLSL. The GPU's are already plenty programmable by design.
"His name was James Damore."