Sun Developing Open Media Stack

Graftweed writes to share that Sun is working on a new open video codec called Open Media Stack (OMS). OMS video will be based on H.26x technology and promises to deliver royalty-free open video. This certainly isn't the first attempt at an open codec, hopefully Sun will decide to add something to the table beyond just their name.

Alas, another flavour

[dlanod]

I think it's great that this is being created, but surely I can't be the only person thinking that there is no way this is going to get any sort of traction outside of niche markets with the other codecs being so entrenched already unless there is some insanely good reason to switch (twice the compression with no noticeable quality changes or something similar)?

Re:

[ChrisMounce]

I fully agree (though I also fully support open formats).

What I'm wondering is what will set this apart from Theora (which was linked to in the Slashdot post). Don't we already have royalty-free video, or is what Sun working on significantly more advanced than the Theora codec?

Re:

[mrsteveman1]

Probably that whole "actually being used" part that tends to cause problems

Re:Alas, another flavour

[Sancho]

Theora is a last-generation codec. It's pretty good, but quality/compression-wise, it's comparable to DivX.

What we really want is something which is comparable to h.264.

Re:Alas, another flavour

[setagllib]

Except that Sun's work is based on h261, because it's so old that no patents can possibly apply to it any more. Dirac is a current/next-generation codec that's also royalty free, and certainly a lot closer to completion than Sun's offering. The FAQ for OMS considers Theora and Dirac as friendly competition, which is fair enough, but really, why not just put more talent into Dirac?

Re:

[makomk]

Dirac is based on wavelets, and I think the entire area of wavelets (being relatively new and shiny) is a total patent minefield. Sure, it's royalty-free now, but if it ever catches on the patent trolls will all come out of the woodwork and it won't stay that way for long.

Re:

[jd]

I'm not convinced it'll go that far, for exactly that reason, unless Sun does something really spectacular. Getting it into web browsers and having Youtube switch, for example. The BBC's Dirac codec is brilliant, but they don't even use their own codec on their own website. They could. They already offer a choice of codecs, one more wouldn't break the bank. They have obviously decided that Dirac just isn't used enough to be worth offering. I don't know what they use on their iPlayer, but I don't think they

Re:

[ardor]

OpenEXR isnt intended for use in the web. Its a HDR format for CG, commonly used for light probes among others. But for the web?

Multimedia

[jd]

If the web is heading in the direction of being an online virtual library and multimedia service, then yes, you want HDR. You want online photographs and mediascapes to be every bit as good as they are in other formats. Web 3.x (or whatever the latest version is under cvs) has chosen to compete with physical publishing, television and cinema. Those are tough fields to compete with, if you're operating at a significantly lower grade.

Re:

[jZnat]

You need to update your working copy of the Web as it is currently under git version control. Use of cvs is so Web 1.0. ;)

Re:

[ardor]

Photographs, sometimes. But the .hdr Format is already popular there, as are several raw formats. These are master copies, and its senseless to display them directly. Use a PNG/JPEG preview instead.

Outside these special cases, HDR has little use. PNG is perfectly ok for diagrams and other synthetic images, JPEG(2000) is perfectly ok for photograph previews and magazine scans.

Re:

[Skinkie]

The problem with dirac was that it was to slow. At least two implementations Schrodinger and the partial not yet finished one in FFMPEG (last SoC) can offer the performance users want to have. Since there is there are now hardware (GPU and FGPA) decoders and encoders for Dirac, a browser plugin going to developed, chances are that this can be a next step in a better codec and free for all. Now I always wondered what will be the audio equivalent of Dirac ;)

Re:Alas, another flavour

[atamido]

First of all, the specification and the reference implementation to produce and read back a valid stream were just finished. A month ago. After a few years of development. And it still doesn't even use all of the features, let alone efficiently.

The reason the BBC isn't using Dirac yet is that it isn't anywhere close to being ready, so it isn't actually usable in any meaningful way. Give it another year of development to get the obvious optimizations done and then the BBC may have a reason to switch to it entirely in the iPlayer. And once the millions of people that use the iPlayer to watch BBC's content prove the value of Dirac, other companies will have an incentive to use it.

Chances are that it will be used in many ways that people won't realize. For instance, Vorbis isn't well known at all in the public, but many game developers use it for audio in games. Game developers love having an open source and royalty free audio decoder with top of the line performance. When Dirac matures, they will love having an open source and royalty free video decoder with top of the line performance too.

Re:

[jd]

The Japanese were looking at a format 30x high-definition, according to a story on Slashdot not too long ago. Dirac might be a reasonable competitor for storing or transmitting a video signal of that magnitude. Another possibility would be to have an all-digital IMAX - regular codecs aren't really up to the job of storing or delivering data of that quality and resolution. A third would be for transmitting video to hand-held devices. I'm not sure ultra-low-power buys you much after a while, but bandwidth to

Re:

[tonyr60]

At least two implementations Schrodinger ....

Presumably the cat is alive in one of the implementations and dead in the other....

Re:

[aproposofwhat]

And once the codebases are merged?

Re:

[orasio]

You got it all wrong. The cat is both alive and dead in both implementations.
Only when the codebases are merged, you can know for sure.
Amateurs.

Re:

[Cyberax]

OGG Vorbis is considered to be one of the best (if not THE best) audio codec.

Re:

[Per Wigren]

Yes, for CD-quality audio it is. The problem is that it's VERY focused at CD/music audio.

It doesn't handle multichannel audio properly. For example it doesn't have a definition of which channel belong to which speaker except for "left" and "right", and it doesn't do multi-channel "joint stereo" (comparing the channels with each other and only keeping what differs, which often is very little) compression.

It also destroys Dolby's analogue surround encoding common in 2-channel movie audio which is used by most

Re:

[Cyberax]

??

http://xiph.org/vorbis/doc/stereo.html [xiph.org] - there is joint stereo support in OGG.

OGG format also has 5.1 support but I have not seen it 'in the wild'.

Re:

[Per Wigren]

Sorry, it seems that I was outdated on the multichannel channel coupling part.

I know it can support 5.1 audio but afaik there is nothing in the spec that defines which channel is center, back left, subwoofer, etc, only front left and right. So if you encode 5.1 audio as OGG/Vorbis the player may play the center channel in the right surround speaker, or similar.

I have no time to check if the spec has been updated in the last 3 years so please correct me if I'm wrong on this topic also. :)

Re:

[Cyberax]

Yes, I can't find it in the spec and there seem to be problems: http://www.hydrogenaudio.org/forums/lofiversion/index.php/t16033.html [hydrogenaudio.org]

But it should not be hard to add, just a small tag with special format describing channel mapping.

Re:

[nine-times]

unless there is some insanely good reason to switch (twice the compression with no noticeable quality changes or something similar)?

IMHO, it's probably a mistake to think that formats really rise or fall on technical merits. It has a lot to do with politics and trust. The two questions are: Does Sun have enough influence to get others to support this format? -and- Do people trust Sun enough to believe that this format will be around for the long haul?

If this new format were suddenly supported out-of-th

Re:Alas, another flavour

[AndrewStephens]

Actually, I think Sun is one of the only companies that could possibly do this. Java is installed on the majority of Windows desktops, and self-updates on each new version. Sun could roll this out as part of a Java update and hardly anyone would notice. Now their only problem is getting content producers to use the codec - good luck with that.

Re:

[tonyr60]

Actually, I think Sun is one of the only companies that could possibly do this. Java is installed on the majority of Windows desktops, and self-updates on each new version. Sun could roll this out as part of a Java update and hardly anyone would notice.

Well if Sun follow the Apple model of including Safari in an application upgrade they are likely to include Solaris in a Java upgrade.

Re:

[Tom9729]

I might be wrong, but I don't think Java updates itself autonomously on Windows. Maybe there's a setting so it will do that, but I think by default it'll only notify/annoy you about the latest version. You (as the user) still have to give it the "go ahead" before it will install anything.

Feel free to mod me down if I'm wrong though.

Re:

[AndrewStephens]

No, you are correct - but I guess most people just click "OK - update" and get on with their lives.

There are very few companies that actually have the power to deploy ubiquitous software - Microsoft (with Widows update), Apple (with iTunes/Quicktime), Sun (with Java update) and maybe Shockwave. Nobody else has the ability to even attempt it.

Re:

[AndrewStephens]

Actually, you would be wrong. Java is installed by the big OEMs, and lots of people install it themselves. Sun used to claim 85-90% or something, which I am not sure I quite believe, but it is at least three quarters in my experience.

Maybe corporate desktops don't have it installed (although they would if the IT department just left the OEM base install on the hard drive), but corporate users aren't going to be downloading a new video codec not matter what anyway.

Re:

[Anonymous Coward]

All they have to do is include it with the various java distributions and suddenly a billion lazy programmers have a free, well documented, easy to use, cross-platform codec that requires no additional tooling. There are libraries out there for using other codecs, but the 2 steps it takes to develop for them (download, then add library to build path) is such a huge barrier to entry they might as well not exist.

Of course they could have just supported one of the other open formats, but why would you blow to

Re:

[Anonymous Coward]

... like their "open source" operating system which is not actually open source.

OpenSolaris is certified Open Source [opensolaris.org] and there are already a half-dozen distributions based on OpenSolaris such as Nexenta [nexenta.org] and Schillix [berlios.de]. If you don't like Sun's management, fork the code and roll your own distro.

Re:

[Sancho]

Oh, to have mod points for you.

Some people just don't understand that "Open Source" doesn't mean "license-compatible with my license of choice."

Just their name

[DennisZeMenace]

> hopefully Sun will decide to add something to
> the table beyond just their name.

The *Java* Sun Open Media Stack ?

Re:

[_merlin]

Different context for the word stack: in this case it's a stack of components, like video and audio codecs sitting on top of some stream format sitting on a container file format. You know - like when they talk about a protocol stack (HTTP on top of TCP on top of IP on top of Ethernet, etc.).

Those of us in EE call it a "system"

[Theatetus]

like, R(s) -> E(s)G(s) -> Y(s) kind of "system".

Re:

[surfingmarmot]

> hopefully Sun will decide to add something to > the table beyond just their name. The *Java* Sun Open Media Stack ?

Erm, That's "the Sun *Java* Open Media Stack". Let's keep those priorities straight. There ain't no Java without the Sun.

I to wanted to say JAVA

[aliquis]

Yeah, this surprised me to, so I searched for JAVA and look what I found!

Anyway.

JAVA short for JAva Video & Audio.

I thought h.26x was patented

[tepples]

I thought there were essential patents without Free licenses [wikipedia.org] on the H.26x technologies, such as the H.264 Advanced Video Coding used in MPEG-4 part 10.

Re:

[dgatwood]

That's nothing. I thought OMS [wikipedia.org] was trademarked.... Maybe Gibson hasn't kept up the registration....

Re:

[trawg]

I just assumed they meant an earlier version than h.264, like h.263 - though Wikipedia tells me US patent law lasts 20 years and h.263 still has a few years left to go, by the looks. Maybe there's another h.26x on which patent has expired, or patent owners have given up rights to it or something.

This is sort of a good example of why Sun's announcement is good news, assuming they're actually going to make it open and not get sued into oblivion for patent infringement. It's such a pain in the ass trying to fi

Huh?

[Watson Ladd]

Isn't there already a gpl'd alternative to .flv [flumotion.net]? What advantages are there in sun's offering? And given that the patent fees on .mp4 are so low, is that really needed?

Re:

[Tom9729]

I think one of the main advantages will be Sun's name behind it.

Re:

[dave420]

That seems to be a gpl'd alternative to Flash Media Server, not FLV. FLVs use H263, VP6, and (in beta) H264 as their video codecs.

Xvid

[TheGreatOrangePeel]

...uhm. not only is Xvid [wikipedia.org] an "attempt" at an open codec, it's arguably a success. I use it for just about all of my encoding, andyway and it's certainly more of a success than Theora.

Re:Xvid

[Cajun Hell]

It's only successful from a technical perspective. The patents keep it underground. If anyone with money tries to use Xvid, they'll either have to license the patents, or they'll be in court. Xvid is useless to Sun and their customers.

Re:

[ne0n]

XviD is technically open source, but not "open" in the ways that matter to Sun's target audience. Same thing with x264 - an incredibly popular codec that's unusable for enterprises unless you license the patents.

The reason Theora exists: it's the least horrible fully-open (beer, speech) codec available.
Keep your eyes on Dirac, someday you'll be able to play Dirac-encoded stuff on a home computer.

Seriously? Why?

[rmdir -r *]

What's wrong with existing solutions? Xiph [xiph.org] has a pretty good container format, and a codec comparable divx/xvid, while the BBC has recently finished Dirac [sourceforge.net], which is not quite ready, but which has the advantage of being:

• Patent and royalty free (the BBC worked very hard at this)
• GPLv2, LGPL, MIT or MPL licensed reference implementation
• Finished: the bitstream has been frozen, etc. Integration with container formats isn't quite there though.
• Better than h.264

So why is trying making a patent-free h.264 clone worth the time? You are certainly duplicating effort, and we already have solutions.

NIH, perhaps? Too many bored engineers?

Re:

[fm6]

What you're really asking is why use H.264 at all? Obviously all the folks who use H.264 have looked at the alternatives you mention and rejected them. I dunno why, but it's not to keep their engineers from getting bored. There a more cost effective solution [opm.gov].

Re:Seriously? Why?

[Jherek Carnelian]

Obviously all the folks who use H.264 have looked at the alternatives you mention and rejected them.

That's an awfully big assumption. From my experience within the corporate world, I'd feel reasonably confident in saying that not even half of the folks who use h.264 even know that dirac exists, much less have looked at it as an alternative to h.264.

Re:Seriously? Why?

[rmdir -r *]

That's an awfully big assumption. From my experience within the corporate world, I'd feel reasonably confident in saying that not even half of the folks who use h.264 even know that dirac exists, much less have looked at it as an alternative to h.264.

It is important to note that Dirac was only finished this year.

Re:

[Wavebreak]

I don't have experience in the corporate world, but I do have personal experience, and I believe you're quite right when it comes to people not having heard about dirac. I personally did hear about it a few years ago, but had no clue that there was an actual implementation out by now, and didn't even remember the name. And, while no pro or even much of an ethusiast, I am generally reasonably well-informed about new codecs and such. While not a representative sample by any means, I'd belive that it'd be fair

Re:Seriously? Why?

[trawg]

Xiph has a pretty good container format, and a codec comparable divx/xvid

Last time I checked the open source Xiph stuff wasn't really comparable to DivX/Xvid - that was a while ago so maybe they've made advances since then, but the last benchmarks I saw showed it was not as good for quality/bitrate.

I have been posting about Dirac in almost every thread on video codecs on Slashdot hoping to raise awareness; the fact that it is truly free and (hopefully) not going to be encumbered by patent rubbish means we might stand a chance of freeing Internet video from the clutches of Adobe/Flash and all the h264/other codec patent holders.

So why is trying making a patent-free h.264 clone worth the time? You are certainly duplicating effort, and we already have solutions.

NIH, perhaps? Too many bored engineers?

My first reaction on reading this was "this is awesome news because the more options the better", which is typically my attitude towards most software. That said, the more people working on Dirac, the better - the BBC have done the hard yards and having a pool of awesome Sun engineers working on it and improving it would certainly help matters.

Re:

[MightyYar]

How do you unseat Flash? Unless the codec is efficient enough to be decoded by Javascript, you won't have anything more commonly installed available.

It's not like people use Flash for the quality, performance, or efficiency! :)

Re:

[True Vox]

FYI, you CAN get good, high quality out of flash. [youtube.com] Of course, YMMV, and it's not EFFICIENT by any stretch of the imagination... ;)

Re:Seriously? Why?

[trawg]

How do you unseat Flash? Unless the codec is efficient enough to be decoded by Javascript, you won't have anything more commonly installed available.

If I had unlimited resources and was hugely philanthropic (and/or just wanted to destroy Flash as the de facto standard for web video), I'd do it like this:

1) develop an open source video codec that is a) comparable in quality/bitrate to mpeg4/h264) and b) not encumbered by patents and does not conflict with existing patents (this is almost certainly the hardest part - even starting from scratch chances are you're going to step on someone's patent portfolio)

2) create an open source player plugin for as many browsers on as many platforms as I could find, with some nice basic functionality and published specs so anyone else could create one for their browser/platform of choice.

3) create open source tools for easy encoding/transcoding of existing content to your content (note that this step might require your transcoding tool to be commercial - in order to do this legitimately I'd say you'd need to buy a license to decode things like mpeg4 into a new format). Publish the shit out of your encoding process and let the open source community make free tools. (This step is, I feel, ridiculously important. Video creation is still a bit of a pain in the ass and unless you can make it easy for people to use it, it'll never take off.)

4) create open source DirectShow filters and all the other crap needed to make your video codec work seamlessly on Windows, and distribute as a simple Windows installer. Make sure they're explicitly redistributable as part of the license and let all those codec pack creators help spread the word.

5) parter with, or create, a site with a bunch of video to a) demonstrate how well it works and b) promote it and help foster adoption. There is an assload of excellent Creative Commons content out there to start with.

(Optional) 6) Create a new company providing commercial services for all of the above for companies that want to go the extra mile (bulk encoding services, streaming and distribution, hosting, etc).

All non-trivial steps!

Re:

[trawg]

None of my systems have ever come with Flash pre-installed. It's always been an add-on (maybe if you buy a Dell or something with a pre-packaged OS it has Flash in it).

Re:

[xenocide2]

A few months ago, I would have agreed that better codecs / plugins provided by browsers by default would be a death blow to flash. Now, I'm inclined to disagree with my previous self. Flash is important for more reasons than just because it can display video poorly. It also provides a way to serve advertisements. Instead of mixing streams or rendering ad overlays into the video, flash can render advertising overlays to the video in real time. It's a wonderful technology for them; it serves ads and is nearl

Re:

[trawg]

Definitely true - but I would argue that rendering text over the top of video in a web-based player is several orders of magnitude easier than creating a new patent-free video codec, so that sort of functionality would be pretty easy to add in - in fact, I'd say that once you had the groundwork of video created, getting people to add in features like that would follow pretty quickly, and you'd possibly have a complete and open Flash competitor in the market even sooner!

Re:

[xenocide2]

A video codec is different than a programmable canvas plugin to a web browser. I'm not sure it's easier, and even if you did have an open flash competitor, have you really won anything significant? Dumb programmers can still make CPU hungry modules, and ads are still in your video. DOH!

Re:

[Fweeky]

And let's not forget Matroska [wikipedia.org], which has done considerably better than Ogg when it comes to video.

Re:

[arose]

Matroska is a container format, not a video codec.

Re:

[xenocide2]

Isn't ogg also a container format?

Re:

[Fweeky]

Yes, and I was replying to "Xiph [xiph.org] has a pretty good container format", hence comparing it to Ogg. I've seen way more .mkv's than I have .ogv and .ogm's, and those I have seen have tended to be more akward to make work than a mkv.

Re:

[TheRaven64]

Theora is based on VP3, which is very old. It's roughly equivalent in terms of quality to H.261. It's completely patent-free, but is still officially in beta (I think - it was at the end of last year) which frightens off a lot of companies. It is also not playable by default for most users.

Dirac is a next-generation CODEC, which aims to be patent-free and is only just nearing a useable state. The processing requirements for Dirac are huge. Encoding it can't be done in anything like real time and dec

Re:

[rmdir -r *]

Dirac is a next-generation CODEC, which aims to be patent-free and is only just nearing a useable state. The processing requirements for Dirac are huge. Encoding it can't be done in anything like real time and decoding requires more CPU than H.264, which limits it to very modern desktops - you won't be seeing it in something like an iPhone for a few years.

The Dirac people claim that if you turn off some advanced features, it performs like h.264, some more, like divx, a few more, like mpeg2. I heard that fr

Oh, they'll add more than just their name

[RalphBNumbers]

They'll add DRM!

This is "derived out of Sun's Open Media Commons initiative", which was in turn based on Project DReaM, which was Sun's attempt at an open source DRM stack.

Why this is important

[Anonymous Coward]

Right now, the working drafts for the HTML 5 specification specify a <video> element, but doesn't specify a codec to go with it. Unfortunately, there's no single video codec which is acceptable to all web browser vendors. Mozilla (and Opera, I think) will not go for something patent-encumbered, while Apple and Microsoft find Theora unacceptable, because of the risk of submarine patents.

Having a modern non-proprietary codec specified which all browser vendors could interoperably implement would give a

If Sun wanted to do something smart

[Dracos]

They'd back Xiph's codecs (Ogg Vorbis for audio, Ogg Theora for video) instead of reinventing the wheel yet again. This is how MS would play the open format game, not someone who can actually be believed when they talk about supporting open, well, anything.

But, this is Sun... they're opposed to making smart moves.

Re:

[Anonymous Coward]

I thought Theora was pretty bad as far as quality - like MPEG1 bad.

Sounds familiar

[Rgb465]

Does this sound an aweful lot like Theora [wikipedia.org] to anyone else?

Re:

[dave420]

Yes, in that they're both open-source video codecs. However Theora's technical abilities are stuck in the late 1990s, whereas Sun's offering hopefully won't be.

Blog entry about OMS Video

[Anonymous Coward]

There is a blog entry [sun.com] for OMS Video.

Sun OMG

[Nushio]

Personally, I'm still waiting for Sun to release their Open Media Games...

Re:

[mhall119]

Personally, I'm still waiting for Sun to release their Open Media Games...

You mean like Wonderland [java.net]?

Re:

[mhall119]

Or maybe Darkstar [java.net]?

Question

[fabu10u$]

How again does Sun make money these days?

Streaming Media Server

[Ambidisastrous]

I was curious too, so I looked on Google Finance [google.com]. Most of their visible activity does seem to be dedicated to blogging and thwarting Microsoft, pretty much like IBM -- big iron and enterprisey stuff. But actually, the financial headlines do a much better job than sun.com at explaining how they make money. Here's a press release from today:
Sun Expands Sun Streaming System to Deliver Industry's Most Scalable and Flexible Video Delivery Over IP Platform [businesswire.com]

This was released today. The releases on this and OMS don'

applets

[slack_prad]

When this is finally done, I just hope it won't need the Java plugin on the browser to view the video.

Java integration

[heroine]

Wonder if Sun is firstly, actually implementing something or just traveling to meetings & specing it, and secondly, trying to write Java support for all its efforts or just acting like a standard, dysfunctional corporation. Having said that, does anyone still care about codecs?

H.26x: what is it?

[ajs]

Nice article on what H.26x is at ddj: http://www.ddj.com/201203492 [ddj.com]

I had no idea how tangled the standards were... ugh.

doom9 has more info..

[SD-Arcadia]

You can check out the progress of Shrodinger (A Dirac implementation) over here: http://forum.doom9.org/showthread.php?t=135176 [doom9.org]