Slashdot Mirror

You know what's also implemented, available and in use? Matroska
While it isn't perfect it has certainly seen more use than Ogg, at least in the video world. Haven't read the article, because it's slashdoted, but I assume it's about the fact that the Ogg container was initially designed as a transport stream format for audio. While that's nice for streaming most video sites use progressive download to deliver the files, so it's not an advantage in that situation. I read that Xiph is only now working on implementing a keyframe index to allow proper seeking in video. Ogg also makes it complicated to integrate support for new codecs. You basically have to rewrite the splitter to support them. See here posts 48 to 52. The last post also has a link to an older hardwarebug article describing problems with the Ogg container.

My Dark Shikari is newer ;-)
Source

Q?: Is Theory still improving much? Is it possible to improve it a lot without breaking backward compatibility, or would that require breaking legacy support?
The encoder still has a good bit of room to improve; it still lacks proper pixel-domain RD optimization and could use some psychovisual improvements as well. It is primarily crippled by the ancient VP3 specification though. In theory it may eventually be able to eke ahead of Xvid, as while Xvid was a very good MPEG-4 ASP encoder, it never had any significant psy optimizations, so even though Theora as a video format is probably slightly worse, the encoder should be able to surpass Xvid with psy optimizations in some cases.
The primary problem for adoption (not for quality) is its lack of features. For many many use-cases, one needs fine-grained control over various parameters, such as the buffer size, the maximum bitrate, the latency, and so forth. The encoder is currently very inflexible, and with no alternative encoding libraries, this leaves very little choice for many applications. For practical purposes, it only has one speed mode: "slow as hell", which makes it basically impossible to stream anything above SD.

That's...an incredible under-statement on your part!

Well, yeah, and I also understand why you are disinclined to being diplomatic. I've been modded troll for posting the comparisons linked in my post here on slashdot and the "It's just as good. really!!!" crowd is getting on my nerves as well. I especially hate the "you don't need to trust that comparison, try encoding yourself" crowd, because they obviously never followed their own advice or they wouldn't make such statements.

x264 started being developed at about the same time the Theora bitstream was frozen and if you look at the progress both projects have made it looks pretty bad for Theora. x264 had at most one or two main developers (concurrently) and a couple of contributors, summer of code projects and a few corporate sponsored improvements and look what they managed to create. A new unique rate control (1; 2), low latency error resilient encoding(3), psy optimizations(4), speed optimizations(5) for multiple processing architectures(6) and it's widely used. Youtube, Facebook, other companies, the guys who post 720p encodes on bittorrent, you name it!
At the same time Theora still has problems doing encodes without losing frames if you want to hit a certain filesize. I like Theora, it's good to have a free alternative when you need one, but Xiph could have done a lot more with it in the time they had and the codec has a really annoying fanclub.

One of the problems Xiph has in my opinion is that they want to do everything. They have a bunch of good codecs (mainly audio) like Vorbis, FLAC, Speex and CELT, but then they develop their own container and subtitle formats instead of just using or adapting the ones that are all ready out there and better than what Xiph has to offer (MKV, ASS/SSA to name two). With Theora they went the adaption route but took a little too long to deliver. They don't even have to innovate. They should just look at x264 (especially mb-tree and the psy optimizations), see how they can adapt it for Theora and port it over and bam! you've got a halfway decent free video codec.

My Dark Shikari is newer ;-)
Source

Q?: Is Theory still improving much? Is it possible to improve it a lot without breaking backward compatibility, or would that require breaking legacy support?
The encoder still has a good bit of room to improve; it still lacks proper pixel-domain RD optimization and could use some psychovisual improvements as well. It is primarily crippled by the ancient VP3 specification though. In theory it may eventually be able to eke ahead of Xvid, as while Xvid was a very good MPEG-4 ASP encoder, it never had any significant psy optimizations, so even though Theora as a video format is probably slightly worse, the encoder should be able to surpass Xvid with psy optimizations in some cases.
The primary problem for adoption (not for quality) is its lack of features. For many many use-cases, one needs fine-grained control over various parameters, such as the buffer size, the maximum bitrate, the latency, and so forth. The encoder is currently very inflexible, and with no alternative encoding libraries, this leaves very little choice for many applications. For practical purposes, it only has one speed mode: "slow as hell", which makes it basically impossible to stream anything above SD.

That's...an incredible under-statement on your part!

Well, yeah, and I also understand why you are disinclined to being diplomatic. I've been modded troll for posting the comparisons linked in my post here on slashdot and the "It's just as good. really!!!" crowd is getting on my nerves as well. I especially hate the "you don't need to trust that comparison, try encoding yourself" crowd, because they obviously never followed their own advice or they wouldn't make such statements.

x264 started being developed at about the same time the Theora bitstream was frozen and if you look at the progress both projects have made it looks pretty bad for Theora. x264 had at most one or two main developers (concurrently) and a couple of contributors, summer of code projects and a few corporate sponsored improvements and look what they managed to create. A new unique rate control (1; 2), low latency error resilient encoding(3), psy optimizations(4), speed optimizations(5) for multiple processing architectures(6) and it's widely used. Youtube, Facebook, other companies, the guys who post 720p encodes on bittorrent, you name it!
At the same time Theora still has problems doing encodes without losing frames if you want to hit a certain filesize. I like Theora, it's good to have a free alternative when you need one, but Xiph could have done a lot more with it in the time they had and the codec has a really annoying fanclub.

One of the problems Xiph has in my opinion is that they want to do everything. They have a bunch of good codecs (mainly audio) like Vorbis, FLAC, Speex and CELT, but then they develop their own container and subtitle formats instead of just using or adapting the ones that are all ready out there and better than what Xiph has to offer (MKV, ASS/SSA to name two). With Theora they went the adaption route but took a little too long to deliver. They don't even have to innovate. They should just look at x264 (especially mb-tree and the psy optimizations), see how they can adapt it for Theora and port it over and bam! you've got a halfway decent free video codec.

My Dark Shikari is newer ;-)
Source

Q?: Is Theory still improving much? Is it possible to improve it a lot without breaking backward compatibility, or would that require breaking legacy support?
The encoder still has a good bit of room to improve; it still lacks proper pixel-domain RD optimization and could use some psychovisual improvements as well. It is primarily crippled by the ancient VP3 specification though. In theory it may eventually be able to eke ahead of Xvid, as while Xvid was a very good MPEG-4 ASP encoder, it never had any significant psy optimizations, so even though Theora as a video format is probably slightly worse, the encoder should be able to surpass Xvid with psy optimizations in some cases.
The primary problem for adoption (not for quality) is its lack of features. For many many use-cases, one needs fine-grained control over various parameters, such as the buffer size, the maximum bitrate, the latency, and so forth. The encoder is currently very inflexible, and with no alternative encoding libraries, this leaves very little choice for many applications. For practical purposes, it only has one speed mode: "slow as hell", which makes it basically impossible to stream anything above SD.

That's...an incredible under-statement on your part!

Well, yeah, and I also understand why you are disinclined to being diplomatic. I've been modded troll for posting the comparisons linked in my post here on slashdot and the "It's just as good. really!!!" crowd is getting on my nerves as well. I especially hate the "you don't need to trust that comparison, try encoding yourself" crowd, because they obviously never followed their own advice or they wouldn't make such statements.

x264 started being developed at about the same time the Theora bitstream was frozen and if you look at the progress both projects have made it looks pretty bad for Theora. x264 had at most one or two main developers (concurrently) and a couple of contributors, summer of code projects and a few corporate sponsored improvements and look what they managed to create. A new unique rate control (1; 2), low latency error resilient encoding(3), psy optimizations(4), speed optimizations(5) for multiple processing architectures(6) and it's widely used. Youtube, Facebook, other companies, the guys who post 720p encodes on bittorrent, you name it!
At the same time Theora still has problems doing encodes without losing frames if you want to hit a certain filesize. I like Theora, it's good to have a free alternative when you need one, but Xiph could have done a lot more with it in the time they had and the codec has a really annoying fanclub.

One of the problems Xiph has in my opinion is that they want to do everything. They have a bunch of good codecs (mainly audio) like Vorbis, FLAC, Speex and CELT, but then they develop their own container and subtitle formats instead of just using or adapting the ones that are all ready out there and better than what Xiph has to offer (MKV, ASS/SSA to name two). With Theora they went the adaption route but took a little too long to deliver. They don't even have to innovate. They should just look at x264 (especially mb-tree and the psy optimizations), see how they can adapt it for Theora and port it over and bam! you've got a halfway decent free video codec.

My Dark Shikari is newer ;-)
Source

Q?: Is Theory still improving much? Is it possible to improve it a lot without breaking backward compatibility, or would that require breaking legacy support?
The encoder still has a good bit of room to improve; it still lacks proper pixel-domain RD optimization and could use some psychovisual improvements as well. It is primarily crippled by the ancient VP3 specification though. In theory it may eventually be able to eke ahead of Xvid, as while Xvid was a very good MPEG-4 ASP encoder, it never had any significant psy optimizations, so even though Theora as a video format is probably slightly worse, the encoder should be able to surpass Xvid with psy optimizations in some cases.
The primary problem for adoption (not for quality) is its lack of features. For many many use-cases, one needs fine-grained control over various parameters, such as the buffer size, the maximum bitrate, the latency, and so forth. The encoder is currently very inflexible, and with no alternative encoding libraries, this leaves very little choice for many applications. For practical purposes, it only has one speed mode: "slow as hell", which makes it basically impossible to stream anything above SD.

That's...an incredible under-statement on your part!

Well, yeah, and I also understand why you are disinclined to being diplomatic. I've been modded troll for posting the comparisons linked in my post here on slashdot and the "It's just as good. really!!!" crowd is getting on my nerves as well. I especially hate the "you don't need to trust that comparison, try encoding yourself" crowd, because they obviously never followed their own advice or they wouldn't make such statements.

x264 started being developed at about the same time the Theora bitstream was frozen and if you look at the progress both projects have made it looks pretty bad for Theora. x264 had at most one or two main developers (concurrently) and a couple of contributors, summer of code projects and a few corporate sponsored improvements and look what they managed to create. A new unique rate control (1; 2), low latency error resilient encoding(3), psy optimizations(4), speed optimizations(5) for multiple processing architectures(6) and it's widely used. Youtube, Facebook, other companies, the guys who post 720p encodes on bittorrent, you name it!
At the same time Theora still has problems doing encodes without losing frames if you want to hit a certain filesize. I like Theora, it's good to have a free alternative when you need one, but Xiph could have done a lot more with it in the time they had and the codec has a really annoying fanclub.

One of the problems Xiph has in my opinion is that they want to do everything. They have a bunch of good codecs (mainly audio) like Vorbis, FLAC, Speex and CELT, but then they develop their own container and subtitle formats instead of just using or adapting the ones that are all ready out there and better than what Xiph has to offer (MKV, ASS/SSA to name two). With Theora they went the adaption route but took a little too long to deliver. They don't even have to innovate. They should just look at x264 (especially mb-tree and the psy optimizations), see how they can adapt it for Theora and port it over and bam! you've got a halfway decent free video codec.

My Dark Shikari is newer ;-)
Source

Q?: Is Theory still improving much? Is it possible to improve it a lot without breaking backward compatibility, or would that require breaking legacy support?
The encoder still has a good bit of room to improve; it still lacks proper pixel-domain RD optimization and could use some psychovisual improvements as well. It is primarily crippled by the ancient VP3 specification though. In theory it may eventually be able to eke ahead of Xvid, as while Xvid was a very good MPEG-4 ASP encoder, it never had any significant psy optimizations, so even though Theora as a video format is probably slightly worse, the encoder should be able to surpass Xvid with psy optimizations in some cases.
The primary problem for adoption (not for quality) is its lack of features. For many many use-cases, one needs fine-grained control over various parameters, such as the buffer size, the maximum bitrate, the latency, and so forth. The encoder is currently very inflexible, and with no alternative encoding libraries, this leaves very little choice for many applications. For practical purposes, it only has one speed mode: "slow as hell", which makes it basically impossible to stream anything above SD.

That's...an incredible under-statement on your part!

Well, yeah, and I also understand why you are disinclined to being diplomatic. I've been modded troll for posting the comparisons linked in my post here on slashdot and the "It's just as good. really!!!" crowd is getting on my nerves as well. I especially hate the "you don't need to trust that comparison, try encoding yourself" crowd, because they obviously never followed their own advice or they wouldn't make such statements.

x264 started being developed at about the same time the Theora bitstream was frozen and if you look at the progress both projects have made it looks pretty bad for Theora. x264 had at most one or two main developers (concurrently) and a couple of contributors, summer of code projects and a few corporate sponsored improvements and look what they managed to create. A new unique rate control (1; 2), low latency error resilient encoding(3), psy optimizations(4), speed optimizations(5) for multiple processing architectures(6) and it's widely used. Youtube, Facebook, other companies, the guys who post 720p encodes on bittorrent, you name it!
At the same time Theora still has problems doing encodes without losing frames if you want to hit a certain filesize. I like Theora, it's good to have a free alternative when you need one, but Xiph could have done a lot more with it in the time they had and the codec has a really annoying fanclub.

One of the problems Xiph has in my opinion is that they want to do everything. They have a bunch of good codecs (mainly audio) like Vorbis, FLAC, Speex and CELT, but then they develop their own container and subtitle formats instead of just using or adapting the ones that are all ready out there and better than what Xiph has to offer (MKV, ASS/SSA to name two). With Theora they went the adaption route but took a little too long to deliver. They don't even have to innovate. They should just look at x264 (especially mb-tree and the psy optimizations), see how they can adapt it for Theora and port it over and bam! you've got a halfway decent free video codec.

That a free codec like Theora exists is good, and it is estimated that Theora can be competitive with or even a bit better than XviD (MPEG4 ASP) when it is more developed and adaptive quantization (in version 1.2) and other psy optimizations are implemented

Not true at all. Every credible codec developer has come down on Theora like a ton of bricks. I used to believe in a lighter touch, but with the hard-core propaganda coming out of Xiph, Wikipedia, and Stallman/FSF, I'm not inclined to be so polite... Everyone who knows anything about video codecs calls it out as a piece of junk. The claims that H.261 is just as good were panned as biased, but aren't far off the evaluations given from every other technical source.

Let's try x264 developer Dark Shikari, who first said Theora might be able to compete with MPEG-4 ASP, but then looked into the details, and came back trashing the technical capabilities of the codec:

Dark Shikari8th May 2009, 20:38
Actually, it's worse than I thought. xiphmont just told me Theora has no MV prediction.

NONE.

Every MV is coded as either 6-bit X and 6-bit Y, or with a global static huffman table. This is worse than MPEG-1.

I retract my statement that Theora can ever get near MPEG-4 ASP. Removing MV prediction from x264, by the way, reduces PSNR by 1db at 500kbps on BlackPearl.

http://forum.doom9.org/archive/index.php/t-146893.html

but currently the encoder has some problems that need to be addressed.

The problems with the VP3 format are fundamental and can't be optimized out. The lousy deblocker that makes things worse, the lack of B-frames, etc, etc. (no time to get into much detail here...) Let's give Mike Melanson the last word here:

What I would like to get across here is that Theora is rather different than most video codecs, in just about every way you can name (no, wait: the base quantization matrix for golden frames is the same as the quantization matrix found in JPEG). As for the idea that most DCT-based codecs are all fundamentally the same, ironically, you can't even count on that with Theora- its DCT is different than the one found in MPEG-1/2/4, H.263, and JPEG (which all use the same DCT).
http://multimedia.cx/eggs/dct-pr/

[...] which is unfortunate considering how long the format has been in development.

That's...an incredible under-statement on your part!

You are right, and there is also the fact that Youtube used a really bad H.264 encoder back then. They have recently switched to x264, so while the SD clips are still encoded in baseline profile that should give them a little boost in quality.

All in all that comparison linked in the summary probably done more to hurt Theora than to help it, because it has given the codec a bad name in the video encoding community. One of the main developers of x264 recently posted about the comparison and Theora in general.
The part concerning Theora having the same quality as H.264:

On the other hand, the Theora devs themselves are strongly against any such claims. gmaxwell asked me to "bonk on the head" anyone who claimed that Theora was nearly as good, as good, or better than H.264, because such claims not only create unreasonable expectations that the devs cannot possibly live up to, but also create the impression that "Theora supporters are liars", which is obviously rather counterproductive to what they are trying to do.

That a free codec like Theora exists is good, and it is estimated that Theora can be competitive with or even a bit better than XviD (MPEG4 ASP) when it is more developed and adaptive quantization (in version 1.2) and other psy optimizations are implemented, but currently the encoder has some problems that need to be addressed. The current Theora encoder is relatively slow. It should be much faster than x264 because of the simpler format, but it currently only has a rather small speed advantage (using the defaults on both encoders; same bitrate) on a single core CPU and it doesn't support multithreading while x264 scales almost linearly with the number of cores. Then there is the fact that the current Theora encoder is dropping frames in 1-pass and 2-pass mode which is unfortunate considering how long the format has been in development.

There are a couple of comparisons between x264 and Theora 1.1 and currently it looks like an up to date x264 revision can give better quality than Theora at half the bitrate with slightly slower (single core CPU) or much faster (multi core CPU) encoding speed using the default settings in both encoders.
Videos; current Theora and old x264 (pre MB-Tree)
Metrics ; on animated content
Screenshots; TV capture and video game footage

You are right, and there is also the fact that Youtube used a really bad H.264 encoder back then. They have recently switched to x264, so while the SD clips are still encoded in baseline profile that should give them a little boost in quality.

All in all that comparison linked in the summary probably done more to hurt Theora than to help it, because it has given the codec a bad name in the video encoding community. One of the main developers of x264 recently posted about the comparison and Theora in general.
The part concerning Theora having the same quality as H.264:

On the other hand, the Theora devs themselves are strongly against any such claims. gmaxwell asked me to "bonk on the head" anyone who claimed that Theora was nearly as good, as good, or better than H.264, because such claims not only create unreasonable expectations that the devs cannot possibly live up to, but also create the impression that "Theora supporters are liars", which is obviously rather counterproductive to what they are trying to do.

That a free codec like Theora exists is good, and it is estimated that Theora can be competitive with or even a bit better than XviD (MPEG4 ASP) when it is more developed and adaptive quantization (in version 1.2) and other psy optimizations are implemented, but currently the encoder has some problems that need to be addressed. The current Theora encoder is relatively slow. It should be much faster than x264 because of the simpler format, but it currently only has a rather small speed advantage (using the defaults on both encoders; same bitrate) on a single core CPU and it doesn't support multithreading while x264 scales almost linearly with the number of cores. Then there is the fact that the current Theora encoder is dropping frames in 1-pass and 2-pass mode which is unfortunate considering how long the format has been in development.

There are a couple of comparisons between x264 and Theora 1.1 and currently it looks like an up to date x264 revision can give better quality than Theora at half the bitrate with slightly slower (single core CPU) or much faster (multi core CPU) encoding speed using the default settings in both encoders.
Videos; current Theora and old x264 (pre MB-Tree)
Metrics ; on animated content
Screenshots; TV capture and video game footage

In the United States, x264 is considered warez because distribution thereof infringes a third party's right.

Really? You should tell Youtube, Facebook, Avail Media and all the other US companies currently using x264 that they are using warez. If you pay the licensing fees you can use x264 however you want. Companies are regularly sponsoring the development of features the want implemented, for instance low latency encoding or streaming improvements. Most private use is probably covered by the less than 100k = free clause in the MPEG licensing agreement and it's not like the MPEG-LA really cares what you use to rip your DVDs.

This seems pretty true. In theory x264 can encode content with very low latency[1], and delivering MPEG-4 from previously encoded files is pretty easy, but my search-fu can't find any ready made solution for streaming using RTSP that doesn't involve paying through the nose for the software---although hacking something together with x264 seems very doable.

I don't know about how easy it is with Theora, but it doesn't really matter since it has had no impact on the mobile device market whatsoever.

[1]: http://x264dev.multimedia.cx/?p=249

Flash supports DRM out of the box, and Silverlight has several features that make it better than Flash. Even its open-source implementation is better.

The results are quite different if you compare Theora to a good H.264 encoder like x264.

Some recent comparisons:
Videos Difference especially noticeable on the "The Island" trailer because of the high motion scenes.
Metrics Animated content. Relatively new.
Metrics By one of the Xiph guys. A little older. Huge lead for x264.

Must be a typo. I think you'll find most seem to be pretty favourable to H.264. Unless that is you could provide a single link that shows a Theora video with higher quality than H.264 at the same bitrate?

I could give you about 10 that show otherwise. here's one

Once you compare Theora to a high quality H.264 encoder it looks pretty bad, unfortunately most people here only seem to know that one comparison done by Xiph themselves against the (relatively crappy) Youtube H.264 encoder with animated content where differences are generally harder to spot. The differences at the same bitrate are obvious in other comparisons.

Here are some more:
Videos Encoded with Theora 1.1 and x264 r1259. The "Island" trailer shows the differences well because there are lots of high motion scenes.
Metrics Animated content. Relatively new.
Metrics By one of the Xiph guys. A little older. Huge lead for H.264.

You got to be kidding. Well, there's a reason why you found it better to post as AC, isn't it?

Ad hominem. Perfect start. I post AC because I don't have an account.

After all, if we play the clips available in that site side by side (which are equivalent in any way) then the only difference you may get will be nothing more than subjective opinions which, I believe, would be go up in smoke if a double blind test as conducted. In fact, if you play the 499kbit/sec comparison side by side you will find the image in the Theora/Vorbis clip to be less grainy and sharper than the h.264 counterpart.

If you view them side by side you'll see the blocky pattern in the Theora encoded clips and you'll see the image degrading into a blocky mess once there is a lot of motion, especially in the "Island" Trailer. I didn't specifically point it out as I figured it would be obvious to anyone watching the clips.

The rest of your comments are just laughable. If we are comparing youtube videos to Theora then it is more than legitimate to compare Theora's codec with youtube's h.264 codec. And your absurd complain about animation clips being used in the comparisson, let's not even delve into the obvious details that it is far easier to legally get your hands on HD videos. The link you provided explicitly states that:

I utilized the Blender Foundation's Big Buck Bunny as my test case because of its clear licensing status, because it's a real world test case, and because I have it available in a lossless format. I am not aware of any reason why this particular clip would favor either Theora or H.264.

While this topic isn't about Youtube you stated in your post that Theora and H.264 provide the same quality. You weren't talking about specific implementations. You didn't mention Youtube and the comparison you linked didn't use youtube for H.264 encoding either. The link I provided doesn't state that at all, it is just a comparison of a film trailer and a shuttle launch done by a slashdot member. Furthermore I didn't say that animated content would favor anyone. I said that animation is easy to compress and therefore makes differences harder to spot. Still they come out quite big SSIM. Hint for the graph, the column saying "x264 baseline" could be used by Youtube without without making use of any features they don't use already.

Video encoding is my hobby. I have tested Theora myself when 1.1 came out and they've had some nice improvements, but nothing that brings them anywhere near a good H.264 encoder like x264, especially not at the low bitrates we are talking about. I don't see a reason why I should put up another comparison when there are plenty available. Even one by the Xiph people themselves shows x264 as an H.264 encoder provides a solid lead over Theora, even with defaults. The defaults used to be very fast settings and were changed some time after this comparison IIRC.

A bit off topic:
Posting AC doesn't automatically make me wrong. Not posting my own comparison doesn't automatically make me wrong either.

There's been an enormous improvement in the Linux scheduler in recent months--in some cases the performance improvements are as high as 60-80% with simple multithreaded apps like video encoders. The instant 2.6.32 comes out officially, expect to start seeing some completely absurd results in stupid "comparisons between Linux distros" like these, where the distros that happened to update to .32 trash the ones that haven't yet.

OK WTF is that all about...

RTMP is the Real Time Messaging Protocol that Adobe has developed for streaming stuff over the Internet.

Red5 is a Free Software (LGPL) implementation of the RTMP.

Cygnal is the Gnash project's RTMP server (also Free Software).

Also see more docs on RTMP on the Gnash wiki, and RTMPE on this other wiki.

... and should I care?

Would you like to have control over the software that you run and use? Are you concerned about your software and/or hardware implementing things like the Broadcast Flag? Do you believe in Free Software because it gives you control over your computer?

If you answered "yes" to any of those questions, then you probably should care, as what's going on right now is making it difficult or impossible for you to run Free Software (or even to pick software) to interact with the RTMP protocol -- a protocol that a given website might require you to use to interact with their media content.

maybe they will finally get video and audio chat working.

It was on the list of projects last year.
http://developer.pidgin.im/wiki/SummerOfCode2008

Also FFmpeg, which developed many of the codecs, is participating
http://wiki.multimedia.cx/index.php?title=FFmpeg_Summer_Of_Code

google has actually helped out a ton for ffmpeg

google is why ffmpeg can now decode wmv3 and real codecs (rv3/4).
more info and list of projects:
http://wiki.multimedia.cx/index.php?title=FFmpeg_Summer_Of_Code

Yes, Android has native class support for multimedia, including H.264.

"The SDK specifies the Media API along with its MediaPlayer and MediaRecorder APIs. According to the AudioEncoder class, audio can be encoded to AMR-NB. The VideoEncoder class specifies H.263, H.264, and MPEG-4 SP." source

Yes, Android has native class support for multimedia, including H.264.

"The SDK specifies the Media API along with its MediaPlayer and MediaRecorder APIs. According to the AudioEncoder class, audio can be encoded to AMR-NB. The VideoEncoder class specifies H.263, H.264, and MPEG-4 SP." source

Graphics/Video PS3 has a powerful graphic processing unit with high speed host connection. The GPU is connected to both HDMI and AV multi interface. Although the GPU is connected directly to CBE, no direct access by guest OSes to the GPU is allowed currently. Video mode/format setting is also the role of AV setting driver. PS3 Linux fb driver calls AV setting driver to setup video modes. Currently X server uses virtual frame buffer to render its image. No hardware acceleration is supported under Linux. See the description above section.

- Linux Kernel Overview

Linux on the PlayStation 3 allows for a huge range of homebrew programs to be developed and is entirely and completely sanctioned by Sony. Although the Cell's performance is more than enough to handle most media requirements or render complex 3D graphics, it does lack the teraflops performance of a contemporary GPU's texture fetching hardware. For this reason many complex games aren't possible on the PlayStation 3 through Linux as access to hardware acceleration in the RSX is restricted by a hypervisor.

-Wikipedia

I'm not a game programmer, but it sure seems to me like having good hardware acceleration is going to make the difference between having to split your main general processing power between game logic and graphics rendering OR having 100% of each to handle the task that it is best suited for. So unless you have some justification for saying that I'm "pulling things from my ass" it seems like your statement doesn't really hold water to the generally available info. I've heard that some hacks are showing some limited success in gaining DMA control over the GPU hardware, but nothing is solid and the method by which access is gained is actively being patched out by Sony in an effort to control the channels by which AAA quality games can get onto their hardware.

x264 is proud to announce that we have four students this year through the Videolan project; they will be working on frametype decision, more efficient inter-macroblock search, better psychovisual optimizations, assembly and profiling improvements, and an interesting tree project that will track the use of data throughout the video stream to maximize the quality of pixels that are referenced the most in future frames.

After 6 months of improvement resulting in two major visual optimizations and a 30% speed increase, we're now going to have an incredibly productive summer. Happy encoding!

Are there ANY success stories?

mike melanson, the vlc developer and now linux flash plugin developer, has created a wiki page with what formats youtube supports...

http://wiki.multimedia.cx/index.php?title=YouTube

also you can create flash video using ffmpeg and mencoder
and play it in vlc and mplayer (both flash1 and flash9/vp6 are supported).

gnash will soon be able to play such videos embedded in youtube....

The Ogg bitstream is also hated by just about everyone who knows anything about media file formats. If you're going to propose a new standard Free media format, please use something like Matroska or NUT instead.
And the answer to your question is: no, Snow and Dirac can't be easily inserted into Ogg. Ogg's inability to handle arbitrary codecs is one of the reasons I don't recommend Ogg.

http://multimedia.cx/eggs/hooking-up-the-smpte-vc- 1-decoder-in-ffmpeg/

The Windows Media specs have been available for anyone to implement (Flip4Mac implements WMV for Macs).

I have a feeling that Flip4Mac is based on MS's reference source code, but that's beside your point.

The sad thing is that while the WM specs have been available, the OSS community didn't bother to create open source codecs on their own

But they did. MPEG-2, MPEG-4, H.264, and recently WMV/VC1 have all been implemented from specs by the community.

The possibility of a cheap publicity stunt notwithstanding, we're still blessed with an open source WMV decoder. It will improve the quality of A/V on Linux, and I can't help but think that's a good thing(TM). For those not knowing what the hell I'm talking about, check out this blog entry about it.

Try again. After SVQ3 got too old (quite a while ago), but before h.264 (just recently) Quicktime was using MPEG-4.

I'm not finding any very *good* sources, but at least see the bottom of this page: http://www.apple.com/quicktime/technologies/mpeg4/

Not to mention numerous MPEG-4/MOV samples exist in the wild.

The libavcodec WMV9 decoder wasn't the one I was talking about. See: http://multimedia.cx/eggs/?p=129

No, it didn't use MPEG-4 video, it used Sorenson Video 1 and 3 (SVQ1/3) for the longest time.

And the only reason Quicktime uses a standard conatiner format now is that MP4 was based on the Quicktime MOV format.

And the FFMpeg implementation (and hence the implementation in MPlayer, Xine, VLC, etc) isn't complete yet. It's improving at a rapid pace, but it's not there yet.

Yet another project deserving more attention.

http://wiki.multimedia.cx/index.php?title=FFmpeg_S ummer_Of_Code

Full list

http://code.google.com/soc/