Fantastic. I'm glad the trolls have taken my earlier advice and made some posts which weren't obviously filled with lies.
Lets break this one down: "not provide better quality than practically any other video format" Depends on how you define other formats. H.261/Mpeg-1? Thoroughly trounced by Theora at all bitrates. Mpeg-2? Again, trounced by Theora. VP6, theora kills it at "web" bitrates.
And this is true without any of the enhancements. The new encoder looks like it will make Theora competitive with MPEG 4 part 10.
You've managed to cite comparisons from 2005. The theora decoder was completely rewritten since then. Or perhaps you've been benchmarking the completely broken implementation in mplayer that doesn't even play all files. Theora has significant lower computational complexity than H.264, and the current decoder delivers that. 1080p theora files play fine on my old laptop, which just chokes and sputters on H.264 files.
And yes, Theora does low bitrate well. Thats why it's well suited for the web.
The overwhelming majority of FLV uses VP6, not h.264. Compare it yourself: At low bitrates even Theora outperforms VP6. (VP6 produces lots of blocking artifacts which Theora doesn't have)
Vorbis has never been widely adopted in the sense intended by the grandparent poster.
Examples would be PNG, GIF (now), JPG... where are the highly proprietary alternatives? Zero traction.
The free thing needs to have very high penetration before it denies air to proprietary competition, but once it's there we should be forever free of having to choose between free and compatible.
Wikipedia isn't a "big video site", but they are an enormous site in general. A small amount of video on Wikipedia still translates into a lot of video in total.
Wikipedia doesn't publish traffic statistics but I wouldn't be surprised if the Wikipedia video traffic were more than the fifth most to the hundredth most popular "video" site combined.
Proof? Prove to me that H.264 doesn't violate any third party patents. Prove that this slashdot AJAX comment interface doesn't violate any patents.
You're asking the wrong question.
I don't know about Unreal. Halo uses Vorbis in Ogg. Then again I can't believe that I'm responding to someone who would even suggest that Ogg has patent problems.
Proving a negative is usually hard. With patents proof is not even possible. (but proving a violation is far more straight forward) What is relevant is the decisions of experienced engineers and attorneys and what we have is experienced engineers and attorneys advising their clients (I.e. Mozilla; Wikimedia) that Theora is okay to use. Meanwhile, can you point to anything more credible than a Slashdot comment saying Theora violates anything specific?
I've never seen a game using Vorbis without Ogg. The only reason I can think of that you'd use Vorbis without Ogg is RTP streaming. But no one is dumb enough to make patent claims about Ogg, because there is pretty much nothing to it.
Theora ships with information on the patent status. Beyond that, perhaps you should have your lawyers call Xiph's lawyers. Legal strategy just isn't something that people post about randomly on the internet.
Youtube is using VP6 (except when you explicitly ask for the high bitrate H264 version), which appears to be substantially lower quality than Theora at the low bitrates they are using.
Care to show an example of *any* MPEG-2 codec out performing the current Theora encoder on a typical web-video 500kbit/sec stream? Forget the new enhanced theora encoder, MPEG-2 can't even match the old crap. Plus mpeg-2 is patented to hell and back, you even have to pay for mpeg-2 decoding in Windows to play DVDs!
Can you cite a *single* example showing Vorbis to be glaringly inferior to AAC? At best the listening tests show AAC to edge out Vorbis only for speech samples at the lowest bitrates (where Xiph has Speex, which blows AAC away for those applications). And no multi-channel? wtf. Vorbis supports 255 channels.
I shouldn't expect better from slashdot, but could you at least find lies that are a bit less obvious.
Ogg high overhead? Okay, Ogg/Vorbis+Theora is something like 1% overhead vs a typical of 0.9% overhead for a movie in AVI. You win there. Then again, OGG provides frequent checksums so that a damaged OGG/Vorbis file will *never* break your speakers and damage your hearing. People who have had the misfortune of hitting a corrupted MP3 in their iPod playlist should be able to appreciate the advantage of this approach. What you consider a fault I consider a feature. Egads, room for design differences exists! who would have thought?
(Minor correction: Theora is only "over a decade" old if we consider its prior life as VP3. Vorbis' public development began over a decade ago, but the 1.0 release was 'only' 6 years ago. The point remains that this software has been conspicuously and publicly available for a very long time, it has been widely distributed. It's created by a group whos mission is to avoid patent infringement, and who would obviously take any reasonable remedy to avoid any known risk. It's not an unknown.
Microsoft got a billion dollar judgment against them (now held on appeal) related to Lucatel submarine patents covering decade old MP3 technology that MSFT was fully paid up on.... So obviously no one is safe from the patent menace but to claim that the Ogg stuff is somehow excessively vulnerable seems somewhat ludicrous considering the publicly available facts.)
I can't find any evidence of Microsoft opposing it on those grounds. Can you point it out in the list archives? It would be rather silly of them to do so since they already ship the Xiph codecs in quite a few products.
As far as Nokia and Apple go... Both are patent holders participating in the MPEG LA pool, both receive fees when non-patent holders use MPEG codecs. Both can avoid paying the same fees themselves by entering into confidential cross-licensing and covenants not to sue with other pool members rather than paying into the pool. (Or alternatively, since we can't actually tell if they are doing that: Both are so large that they would exceed the annual licensing fee caps by a healthy margin.... supporting Theora/Vorbis would allow their smaller competition to save money but not them).
Regardless of the claimed justification this is exactly the sort of result you'd expect when you include parties with clear conflicts of interest in decision making processes.
Considering that Vorbis and Theora have been publicly available for over a decade, distributed in the millions or tens of millions by large and small groups alike, and never resulted in litigation or even public disclosure of claimed infringed patents.... The obvious explanation here is that Apple and Nokia's position is driven not by a desire to avoid infringement but instead by a desire to preserve their vendor lock-in and multimedia-tax income.
(I have no doubt that Nokia has some obscure patent whos 23rd independent claim purports to patent the notion of compressing audio or the like... but such clams would be quickly struck by the USPTO if they were made available for review, thus the cloak-and-daggers game)
Just a tip: You should at least make up some facts to support assertions like that... otherwise their absence is conspicuous.
In this case, the conspicuousness is well justified: The facts don't support you. I can't only assume you're talking about green-marker sniffing morons when you say audiophile, because every objective double-blind listening test on the internet over 64kbit/sec shows Vorbis either clearly beating or tying with the competition.
AAC most certainly is "propritary" by the usual language used on this forum. It is extensively patented. Software and hardware makers have to pay licensing and media distributors are subject to per-use fees. It is impossible to legally use AAC in many countries without fees being paid to the licensing pool.
It's true that Apple didn't play an important role in the creation of AAC (although Nokia did), but like Nokia Apple is a participant in the MPEG LA licensing pool and as such they receive fees for MPEG systems licensing and can avoid paying fees themselves by cross-licensing with other pool members rather than paying into the pool.
It's technically possible to wrap Vorbis up in DRM, and people have done so... The popular DRM system are just wrappers and could neutrally be applied to any codec.
Why do you need JPEG / PNG support in the browser?... So it actually *works* for most users.
The whole "sit back and let a plugin handle it" approach has resulted in a pretty good chunk of the web locked up behind proprietary players and authoring tools. Since it's an explicit goal of the W3C to not propritarize the web by adopting non-royality-free technology some folks argue that they ought not to propritarize the web through inaction either!... and a browser that can't support the basic functionality that people expect while browsing without a bunch of extra plugins isn't doing its job very well.
The HTML5 spec originally specified that, as a baseline, conforming implementations should include a minimum of Vorbis and Theora.
This would mean that web developers would have a reasonable baseline they could target that would work for all users, but still offer up 'higher quality' versions in more efficient alternative formats if the user had the right software.
Sadly, some of the MPEG video patent holders have big voices in the W3C and demanded that there be no baseline. (What a shock: they don't want to have have a more level compatibility playing field because they don't want to have to compete on quality and price).
W3C pulled the baseline due to those demands... but at least they didn't mandate useless or proprietary codecs.
No one proprietary format can gain universal adoption because some companies are always going to push their own, which is why we have this morass of incompatibility... FLV, WMV, Real, ugh. Apple pay Microsoft for a video format? Not if they can help it!
Companies like Apple are perfectly happy having their own walled gardens of incompatible formats since they've made quite a business out of it. The lack of a good standard suits them just fine.
So... providing good working web video becomes a numbers game and it's all up to us users to set things straight by making good choices, which is why this is such big news. Internet standards... protocols, formats, etc. should belong to the public. Anything less will make us perpetual victims to fighting between big companies and leave us subject to constant taxes on our internet use.
There *are* some other codecs which are unencumbered: Motion-jpeg, and H.261 for example. But at web acceptable bit-rates (say between 100 and 600kbit/sec) these codes hardly produce recognizable images for common frame sizes. They just aren't useful. You can make a good argument that Theora is behind the state of the art, but for web use the other free stuff isn't even comparable.
And MPEG3? We should use a dead, patent encumbered, standard for HDTV that is designed for 25+mbit/sec for web use? Give me a break!
Many of the codecs people think are "free" are really quite expensive with per unit encoder, decoder, and encoded media costs. It's easy to ignore these when they are packed up as part of the "Microsoft tax" but their burden on content creators and society in general is pretty substantial.
When you're a Wikipedia, serving hundreds of millions of users per month on donations, this matters. Especially since a key part of their mission is making sure that everyone has the freedom to modify their works without paying tribute to middlemen like Apple and Nokia.
Ah, nothing like Slashdot to bring out the best in humanity. The doom9 comparison is four years old... that would be like comparing something to the MPEG reference code.
The latest work on Theora shows a pretty clear doubling of quality per bitrate vs theora from a few months ago... but since this is Slashdot, I'm sure that little details like that won't slow anyone down. Good job, Nokia.
Have you seen the video conversion instructions on Wikimedia commons?
They appear to include instructions that cover all of your complaints, including rotation. If those instructions are lacking... whats that Wikipedia motto? You can edit?
Your Wikipedia userpage says you're a PHD in computer engineering? I suspect hat "you can edit" also applies to ffmpeg2theora.:)
Good points though!
I might claim that this event is unimportant due to Theora's quality compared to the leading-edge codecs, but it looks like that has been fixed, or soon will be. Obviously no one sane will knock Vorbis' quality.
With the way things are going this sounds like it's going to be quite a fight between the proprietary and open worlds. I can't think of anyone better than Noikia and Apple to play the side of proprietary.... Not even Microsoft seems to be able to pull off, well, evil as completely as those two these days. And with Mozilla and Wikipedia on the other side it's not like either side of this fight is hopelessly out-gunned.
Of course, this is interesting to more than just Wikipedia, but few other players are both as important and have such a clear long-term vision.
Slashdot Mirror
Fantastic. I'm glad the trolls have taken my earlier advice and made some posts which weren't obviously filled with lies.
Lets break this one down: "not provide better quality than practically any other video format" Depends on how you define other formats. H.261/Mpeg-1? Thoroughly trounced by Theora at all bitrates. Mpeg-2? Again, trounced by Theora. VP6, theora kills it at "web" bitrates.
And this is true without any of the enhancements. The new encoder looks like it will make Theora competitive with MPEG 4 part 10.
You've managed to cite comparisons from 2005. The theora decoder was completely rewritten since then. Or perhaps you've been benchmarking the completely broken implementation in mplayer that doesn't even play all files. Theora has significant lower computational complexity than H.264, and the current decoder delivers that. 1080p theora files play fine on my old laptop, which just chokes and sputters on H.264 files.
And yes, Theora does low bitrate well. Thats why it's well suited for the web.
The overwhelming majority of FLV uses VP6, not h.264. Compare it yourself: At low bitrates even Theora outperforms VP6. (VP6 produces lots of blocking artifacts which Theora doesn't have)
Vorbis has never been widely adopted in the sense intended by the grandparent poster.
Examples would be PNG, GIF (now), JPG ... where are the highly proprietary alternatives? Zero traction.
The free thing needs to have very high penetration before it denies air to proprietary competition, but once it's there we should be forever free of having to choose between free and compatible.
No, thats kind of the point. You can download the example clips encoded with the development encoder and play them on your own system if you like.
Wikipedia isn't a "big video site", but they are an enormous site in general. A small amount of video on Wikipedia still translates into a lot of video in total.
Wikipedia doesn't publish traffic statistics but I wouldn't be surprised if the Wikipedia video traffic were more than the fifth most to the hundredth most popular "video" site combined.
Wikipedia uses a Java decoder for Theora (because no browsers include Theora yet) and for Dirac the Java version is incomplete and not real-time.
Proof? Prove to me that H.264 doesn't violate any third party patents. Prove that this slashdot AJAX comment interface doesn't violate any patents.
You're asking the wrong question.
I don't know about Unreal. Halo uses Vorbis in Ogg. Then again I can't believe that I'm responding to someone who would even suggest that Ogg has patent problems.
Proving a negative is usually hard. With patents proof is not even possible. (but proving a violation is far more straight forward) What is relevant is the decisions of experienced engineers and attorneys and what we have is experienced engineers and attorneys advising their clients (I.e. Mozilla; Wikimedia) that Theora is okay to use. Meanwhile, can you point to anything more credible than a Slashdot comment saying Theora violates anything specific?
I've never seen a game using Vorbis without Ogg. The only reason I can think of that you'd use Vorbis without Ogg is RTP streaming. But no one is dumb enough to make patent claims about Ogg, because there is pretty much nothing to it.
Theora ships with information on the patent status. Beyond that, perhaps you should have your lawyers call Xiph's lawyers. Legal strategy just isn't something that people post about randomly on the internet.
Youtube is using VP6 (except when you explicitly ask for the high bitrate H264 version), which appears to be substantially lower quality than Theora at the low bitrates they are using.
As Wikipedia would say: "Citation needed".
Care to show an example of *any* MPEG-2 codec out performing the current Theora encoder on a typical web-video 500kbit/sec stream? Forget the new enhanced theora encoder, MPEG-2 can't even match the old crap. Plus mpeg-2 is patented to hell and back, you even have to pay for mpeg-2 decoding in Windows to play DVDs!
Can you cite a *single* example showing Vorbis to be glaringly inferior to AAC? At best the listening tests show AAC to edge out Vorbis only for speech samples at the lowest bitrates (where Xiph has Speex, which blows AAC away for those applications). And no multi-channel? wtf. Vorbis supports 255 channels.
I shouldn't expect better from slashdot, but could you at least find lies that are a bit less obvious.
Ogg high overhead? Okay, Ogg/Vorbis+Theora is something like 1% overhead vs a typical of 0.9% overhead for a movie in AVI. You win there. Then again, OGG provides frequent checksums so that a damaged OGG/Vorbis file will *never* break your speakers and damage your hearing. People who have had the misfortune of hitting a corrupted MP3 in their iPod playlist should be able to appreciate the advantage of this approach. What you consider a fault I consider a feature. Egads, room for design differences exists! who would have thought?
(Minor correction: Theora is only "over a decade" old if we consider its prior life as VP3. Vorbis' public development began over a decade ago, but the 1.0 release was 'only' 6 years ago. The point remains that this software has been conspicuously and publicly available for a very long time, it has been widely distributed. It's created by a group whos mission is to avoid patent infringement, and who would obviously take any reasonable remedy to avoid any known risk. It's not an unknown.
Microsoft got a billion dollar judgment against them (now held on appeal) related to Lucatel submarine patents covering decade old MP3 technology that MSFT was fully paid up on. ... So obviously no one is safe from the patent menace but to claim that the Ogg stuff is somehow excessively vulnerable seems somewhat ludicrous considering the publicly available facts.)
I can't find any evidence of Microsoft opposing it on those grounds. Can you point it out in the list archives? It would be rather silly of them to do so since they already ship the Xiph codecs in quite a few products.
As far as Nokia and Apple go ... Both are patent holders participating in the MPEG LA pool, both receive fees when non-patent holders use MPEG codecs. Both can avoid paying the same fees themselves by entering into confidential cross-licensing and covenants not to sue with other pool members rather than paying into the pool. (Or alternatively, since we can't actually tell if they are doing that: Both are so large that they would exceed the annual licensing fee caps by a healthy margin. ... supporting Theora/Vorbis would allow their smaller competition to save money but not them).
Regardless of the claimed justification this is exactly the sort of result you'd expect when you include parties with clear conflicts of interest in decision making processes.
Considering that Vorbis and Theora have been publicly available for over a decade, distributed in the millions or tens of millions by large and small groups alike, and never resulted in litigation or even public disclosure of claimed infringed patents. ... The obvious explanation here is that Apple and Nokia's position is driven not by a desire to avoid infringement but instead by a desire to preserve their vendor lock-in and multimedia-tax income.
(I have no doubt that Nokia has some obscure patent whos 23rd independent claim purports to patent the notion of compressing audio or the like... but such clams would be quickly struck by the USPTO if they were made available for review, thus the cloak-and-daggers game)
Just a tip: You should at least make up some facts to support assertions like that... otherwise their absence is conspicuous.
In this case, the conspicuousness is well justified: The facts don't support you. I can't only assume you're talking about green-marker sniffing morons when you say audiophile, because every objective double-blind listening test on the internet over 64kbit/sec shows Vorbis either clearly beating or tying with the competition.
AAC most certainly is "propritary" by the usual language used on this forum. It is extensively patented. Software and hardware makers have to pay licensing and media distributors are subject to per-use fees. It is impossible to legally use AAC in many countries without fees being paid to the licensing pool.
It's true that Apple didn't play an important role in the creation of AAC (although Nokia did), but like Nokia Apple is a participant in the MPEG LA licensing pool and as such they receive fees for MPEG systems licensing and can avoid paying fees themselves by cross-licensing with other pool members rather than paying into the pool.
It's technically possible to wrap Vorbis up in DRM, and people have done so... The popular DRM system are just wrappers and could neutrally be applied to any codec.
Why do you need JPEG / PNG support in the browser? ... So it actually *works* for most users.
The whole "sit back and let a plugin handle it" approach has resulted in a pretty good chunk of the web locked up behind proprietary players and authoring tools. Since it's an explicit goal of the W3C to not propritarize the web by adopting non-royality-free technology some folks argue that they ought not to propritarize the web through inaction either! ... and a browser that can't support the basic functionality that people expect while browsing without a bunch of extra plugins isn't doing its job very well.
The Theora decoder is already integer-only!
The HTML5 spec originally specified that, as a baseline, conforming implementations should include a minimum of Vorbis and Theora.
This would mean that web developers would have a reasonable baseline they could target that would work for all users, but still offer up 'higher quality' versions in more efficient alternative formats if the user had the right software.
Sadly, some of the MPEG video patent holders have big voices in the W3C and demanded that there be no baseline. (What a shock: they don't want to have have a more level compatibility playing field because they don't want to have to compete on quality and price).
W3C pulled the baseline due to those demands... but at least they didn't mandate useless or proprietary codecs.
No one proprietary format can gain universal adoption because some companies are always going to push their own, which is why we have this morass of incompatibility... FLV, WMV, Real, ugh. Apple pay Microsoft for a video format? Not if they can help it!
Companies like Apple are perfectly happy having their own walled gardens of incompatible formats since they've made quite a business out of it. The lack of a good standard suits them just fine.
So... providing good working web video becomes a numbers game and it's all up to us users to set things straight by making good choices, which is why this is such big news. Internet standards... protocols, formats, etc. should belong to the public. Anything less will make us perpetual victims to fighting between big companies and leave us subject to constant taxes on our internet use.
And MPEG3? We should use a dead, patent encumbered, standard for HDTV that is designed for 25+mbit/sec for web use? Give me a break!
Many of the codecs people think are "free" are really quite expensive with per unit encoder, decoder, and encoded media costs. It's easy to ignore these when they are packed up as part of the "Microsoft tax" but their burden on content creators and society in general is pretty substantial.
When you're a Wikipedia, serving hundreds of millions of users per month on donations, this matters. Especially since a key part of their mission is making sure that everyone has the freedom to modify their works without paying tribute to middlemen like Apple and Nokia.
Ah, nothing like Slashdot to bring out the best in humanity. The doom9 comparison is four years old... that would be like comparing something to the MPEG reference code. The latest work on Theora shows a pretty clear doubling of quality per bitrate vs theora from a few months ago... but since this is Slashdot, I'm sure that little details like that won't slow anyone down. Good job, Nokia.
Have you seen the video conversion instructions on Wikimedia commons? They appear to include instructions that cover all of your complaints, including rotation. If those instructions are lacking ... whats that Wikipedia motto? You can edit?
Your Wikipedia userpage says you're a PHD in computer engineering? I suspect hat "you can edit" also applies to ffmpeg2theora. :)
Good points though!
With the way things are going this sounds like it's going to be quite a fight between the proprietary and open worlds. I can't think of anyone better than Noikia and Apple to play the side of proprietary. ... Not even Microsoft seems to be able to pull off, well, evil as completely as those two these days. And with Mozilla and Wikipedia on the other side it's not like either side of this fight is hopelessly out-gunned.
Of course, this is interesting to more than just Wikipedia, but few other players are both as important and have such a clear long-term vision.
Round TWO! FIGHT!