We here at OSAlert have taken somewhat of an interest in the new HTML5 video and audio tags, which should – some day – make embedding audio and video material into web pages as easy and straightforward as embedding images, allowing the web to finally remove the shackles of dreadful Flash video. Sadly, the problem with these new tags are the codecs; as it turns out, browser makers have not reached an agreement about what codecs to choose for video, with mostly Apple throwing a spanner in the works, and Microsoft shining in absence.
The HTML5 video tag obviously requires a standardised codec in order to achieve full interoperability, but the browser makers have been unable to reach a consensus about which codec to implement in their browsers, HTML5 specification editor Ian Hickson writes.
Google implements both Ogg Theora and H.264 in Chrome, but sadly, the expensive licensing costs for H.264 prevents other Chromium distributors from implementing H.264. Google also said that Ogg Theora’s quality-per-bit is not yet good enough for the volume handled by YouTube. Mozilla implements Ogg Theora, but not H.264 due to not being able to provide a license for downstream distributors. Opera will most likely implement Ogg Theora as well, but won’t implement H.264 due to its prohibitive licensing costs.
This means that three of the major browser makers standardise on Ogg Theora. While Theora may not be up to par with H.264 yet, the fact that so many browser makers decided to implement it may speed up its development.
So, this leaves only Apple and Microsoft. Apple has said that they will not implement Ogg Theora support in QuickTime (what Safari uses), citing a lack of hardware support and an uncertain patent situation. This seems like a cop-out on Apple’s end – the other browser makers seem to have no problems with Theora’s patent situation. Microsoft, by the way, hasn’t said anything about this situation at all – it’s not even known when or if they’re going to implement the video tag at all.
According to Hickson, this situation could go either of two ways (or both):
- Ogg Theora encoders continue to improve. Off-the-shelf hardware Ogg Theora decoder chips become available. Google ships support for the codec for long enough without getting sued that Apple’s concern regarding submarine patents is reduced. => Theora becomes the de facto codec for the Web.
- The remaining H.264 baseline patents owned by companies who are not willing to license them royalty-free expire, leading to H.264 support being available without license fees. => H.264 becomes the de facto codec for the Web.
As for audio, no common codec has been required, because there are so many more available. Audio is also of less importance, so Hickson argues that we should just wait and see if any common codec surfaces in the future, at which point it could be made a requirement.
I think once Wikipedia has Ogg content the browser share of browsers that support it natively will increase.
I am quite optimistic because there will soon be a few sites that require ogg.
The web evolves.
The #1 hurdle looking to hold back adoption of Ogg Vorbis/Theora is Google/YouTube’s current preference towards H.264/AAC. Their current HTML5 test page only works in Chrome as it does not use Ogg [ http://youtube.com/html5 ].
I assume the primary reason behind this isn’t really that there’s that noticeable a quality/size difference, but rather almost all the video on YouTube are already encoded with Mpeg4 as Flash and their iPhone app support it. To support Ogg will require a major re-encoding of all their videos meaning quite a great deal of time & money they aren’t prepared to spend right now. If YouTube sticks with MP4 only, then Ogg probably won’t ever get a real shot at taking off :/
Hmm, Google cares for money. And youtube already costs them like 0.5 billion a year. Mostly bandwith I guess.
Ogg Theora is 15% bigger than mp4.
That is a lot of cash.
YouTube did not get where it is based on quality. Google are being a snooty Vimeo wannabe. If they switched everything to OGG over night, the punters would never notice.
The quality of most source material is terrible enough – so Theora and mp4 at the same bitrate would look totally similar.
Resource-wise probably the biggest hurdle for Google/Youtube is the need to re-encode all archive to Theora.
How much does h264 licensing cost for serving that content? Are Google big enough to not have to pay?
What about an upstart video hoster, can they afford MPEG-LA fees?
From 2011 onwards, the owners of the H264 patent have stated that they will charge a fee for transmissions of h264-encoded video streams.
For digital TV stations and websites that have the odd video file here and there, that won’t amount to much, but for a video website like YouTube this will cost a lot more than bandwidth.
Besides, your data is out of date. The Theora encoder is improving apace, and has almost caught up with H264. Besides, most of the existing video files on a site like YouTube are actually encoded in h263, and Theora uses currently LESS bandwidth that h263 for the same quality.
Those bandwidth costs have calculated the wrong way, have a look at this:
http://www.ramprate.com/pdf/RRMarketCommentary-GoogleandYouTube.pdf
Uhm… No. Youtube and most flash videos are not in fact Mpeg4, it’s On2 Technologies VP6, H.263 or H.264 has been since 2004.
As reported by Media player classic
FLV1 == VP6
FLVC == H.263
The newest iteration supports H.264, and while technically flv supports MP4 (FL4) nobody I’m aware of is actually using it that way.
The H.264 support is what really blew people’s skirts up lately, since that’s HOW YT has implemented that little ‘HD’ button.
Edited 2009-07-03 00:44 UTC
Wrong. FLV1 = H.263 (Sorenson Spark). YouTube has always been using H.263 and now it’s using H.264. Old videos on YouTube are encoded in FLV/H.263, new videos on YouTube are encoded in H.264, and the container is sometimes FLV (H.264 in FLV is perfectly normal), sometimes MP4.
AFAIK, only the “watch in HD” streams are encoded in h264. Everything else on YouTube is encoded in either H263 or VP6, both of which are outdone by Theora in its current state of play.
Also, AFAIK, YouTube is actively looking for a more viable solution than H264, because after 2010 it will cost YouTube (as it will anyone else) a small amount of money for every time they send any H264 stream to any recipient.
For a site like YouTube, that could be a lot of transmissions of a lot of videos. Even a small charge for each transmission would end up costing YouTube a fortune by the time it is all added up. This is, AFAIK, exactly why YouTube are looking now to use something else for their codec.
Everything I have downloaded from YouTube in the last couple of months is encoded in H.264. Including the low-res videos. All old videos were encoded in H.263. I have never seen a single video on YouTube encoded in VP6. That’s just an old myth, an old mistake spread by someone and repeated by others. BTW, VP6 is much better than H.263, and probably still better than Theora (which is based on VP3 – sure, the encoder may be improving dramatically, but still, they have to cope with the limitations of a 1990’s generation format, so there’s only so much they can do).
But I don’t see what the H.264 replacement could be. Theora? Will it ever be efficient enough? It may be free to use, but bandwidth isn’t free either.
Even if the current state of play remains static, and no further improvements to Theora encoding (beyond what has already been achieved by Thusnelda) can be made, then it comes down to a trade-off.
The cost of using H264, after 2010, will be a small fee for each transmission of any h264 video file (in addition to a larger license fee for encoding which AFAIK already applies). For a website which features a large collection of videos, the total fee from serving this collection to a large audience this could easily reach prohibitive proportions.
The cost of using Theora to get the same quality as H264 will be a larger filesize. At this time, it is about 20% or so larger. This would cost more in bandwidth (if it cannot be improved upon).
Even at this time the longer-term ongoing cost of the second option would appear to be considerably less.
Edited 2009-07-03 06:18 UTC
And what people seem to ignore completely is that FLV is a CONTAINER, much like MP4 or AVI. The codec is simply the format of the content of the file. You can implement a video player plugin that handles FLV without going near to Flash… indeed Quicktime, as an example, will happily play FLV files, much like it’ll also play 3GP files from my Sony Ericsson mobile phone – so long as it has the codec. VLC is much the same too.
Accepting what is already a bad solution, patent-wise, and complaining that another one is uncertain, patent-wise, is just plain stupid. I think the author hit the nail right on the head.
It’s almost enough to make me want to delete Safari from my Mac Mini and use Firefox 3.5 for everything instead. Oh wait, I already DO use Firefox on my Mac Mini! We’ll just have to see about killing Safari in the future.
Ditto…there’s no reason Apple couldn’t support both Ogg and Mpeg4, and then if some hidden patent were to arise they could just pull Ogg support and push out an update within a matter of days. Problem solved. Their reasoning is complete and utter b.s.
Apple is blatantly lying about the Ogg patent thing. They’ve made their investment in h.264, and they’re going to milk it for all it’s worth.
+1
I said this elsewhere, but related to this exact problem:
—
As I see it, the codec issue can only stay as it is, or get worse:^aEUR”
Apple will never ever give up H.264. Since the introduction of it in Tiger, they have staked their entire media platform on H.264^aEURTMs quality. iTunes / iPod / iPhone / QuickTime / AppleTV. We will always have to supply at least one OGG and one MPEG-4 video file, I don^aEURTMt see that ever going to just one unfortunately.
What worries me most is what Microsoft are going to do if they ever decide to join the party with IE9. They have a number of options:
1. Dictate: Only WMV supported; Microsoft have invested a lot in WMV and it would seem unnatural for them to adopt OGG. Going with WMV-only would prevent IE working with all current Video tags on the net as well as massively peeing-off web developers for having to support a third encode.
2. Tag along: Only MPEG-4. It^aEURTMs common enough and it would work with all existing HTML5 video content on the web.
3 Somebody Else^aEURTMs Problem: Just embed WMP and allow whatever codecs are installed. This is the most likely situation IMO. It would be the least amount of work for them, it would be compatible with existing content and they wouldn^aEURTMt be taken to task for choosing one codec over another. They could easily spear-head WMV as well, by encouraging use of WMV content and using it on their own sites, knowing that despite caliming that they ^aEUR~support the standard^aEURTM HTML5 video tag, in reality much content would be tied to Microsoft^aEURTMs WMV, requiring users to install the necessary codecs for their systems. This woud also stymie general OGG adoption.
—
As it stands however, Firefox 3.5^aEURTMs marketshare will surpass Safari by a wide margin, and it only accepts OGG. Given the choice of OGG or H.264, Chrome will pick OGG, and Opera will support OGG.
Therefore, those using HTML5 video have no choice but to support OGG because quite simply Firefox^aEURTMs marketshare is too great to be ignored. Numbers working for the greater good for once.
That’s really ironic, because Quicktime has the worst H.264 implementation available. Hopefully that will improve with Quicktime X.
But I think you are right regarding Apple and H.264. They will never switch.
I agree with you that Microsoft would likely implement HTML 5 video tags in IE by simply playing the video in a WMP plugin and directing it to play the video. If the computer has the a DirectShow filter installed for the required codec, then it’ll play, end of story. If not, then either display an error or allow the WMP plugin to try to download the codec on the fly (the WMP stand alone player already has this ability, though I’ve never seen it work reliably (I think I’ve seen it work for DivX). I’m pretty sure that DirectShow filters exist for Ogg.
Besides that, I’m not sure Microsoft cares all that much about pushing WMV. Silverlight 1 and 2 supported only VC-1 (the SMTPE standard for “WMV”, of which WMV (including WMV-HD) is Microsoft’s implementation). But for Silverlight 3, they’re adding support for H.264 because lots of content creators have been using it.
So for HTML 5 video tags, I can imagine that they’d be open to supporting something besides WMV since they’ve already shown a willingness to do that for Silverlight 3. Microsoft will support whatever codecs a large percentage of content creators decide to use. If Ogg gets massive support by content creators, then I could see Microsoft shipping a DirectShow filter for Ogg with IE9 or WMP or Windows itself, then go ahead and invoke a WMP plugin to handle HTML 5 video as described above. I don’t think Microsoft is as wedded to WMV as Apple is to H.264.
Anyway, I see Flash as the defacto standard for embedded video for the forseeable future. If Google converted YouTube to using HTML 5 video tags, that would change overnight, but I foresee Adobe making deals with Google to ensure that doesn’t happen. Anyway, Flash provides UI elements that can’t be done with HTML 5, so YouTube will require Flash anyway (whether the Flash plugin is streaming MPEG-4, H.264, .flv, Ogg, or whatever).
I think the history of embedded video in web pages goes like this:
1. Real plugins dominated
2. then QuickTime plugins dominated
3. then WMV plugins dominated
4. now Flash plugins dominate
5. Future: HTML 5 video tags (streaming Ogg???) will be used unless rich UI is required, in which case Flash plugins will dominate. Pages will never use QuickTime plugins except for Apple.com, Silverlight will be on the outskirts in a very small number of high profile pages, WMV plugin use will die off completely, joining Real in the grave.
I’m just randomly speculating.
No. Basic controls handled by the browser, additional controls by JS. No Flash required.
Javascript is no replacement for Flash. It’s horribly slow in comparison, as is the DHTML you’d need to replicate the Flash UI. There is a programming language called OpenLazlo http://en.wikipedia.org/wiki/OpenLaszlo that allows compilation to either Flash or DHTML as a back end and the DHTML loads like treacle. Maybe that is their implementation, but it was totally unusable on most modest desktop machines when I evaluated it.
Sigh!
ECMAScript running under a JIT compiler (such as Squirrelfish or tracemonkey) is not horribly slow at all.
I think you must be getting confused by the horrible slow performance of the non-compliant Javascript engine in IE.
YouTube can easily provide UI elements in conjunction with HTML 5 by using W3C standards CSS3, SVG, animated PNG and ECMAScript. Most browser applications support all of this just fine.
http://arstechnica.com/open-source/news/2009/05/google-dailymotion-…
http://www.dailymotion.com/openvideodemo
Flash is no longer required.
No. Rich UIs are easily possible as long as browsers, in addition to the HTML 5 tag and Ogg codecs, also support current W3C standards that have been stable now for about 5 years or in some cases longer. The set of stable standards defined in the acid3 compliance tests are perfectly adequate for this tsak. There are quite a number of very impressive demonstrations of this technology if you look around the web.
Fortunately, almost all of the different current web browsers that are available support all of this, or will do so very soon (to over 90% compliance, which is good enough). There is only one laggard, really.
PS: I have heard rumours that there are open source projects underway to create ActiveX components that will provide HTML 5 and other W3C standards support to that one recalcitrant web browser also, so even that holdout browser may not prove to be a problem.
Edited 2009-07-03 00:15 UTC
I just wanted to say that the portable player I use is a Meizu M6, just because it had a good sound quality (according to reviews I had read) and because it supports ogg vorbis. If it hadn’t had it, I wouldn’t have bought it. Hopefuly ogg support and knowledge by people will grow over time.
Also, in my experience, ogg vorbis’ quality is superior to mp3’s. Both by having a smaller footprint for the same quality and a better sound quality for files having aprox. the same size.
*WHY* is everyone pushing a technology that is obviously inferior?
Ogg Theory has been proven, time and time again, to have lower quality per unit of data than other codecs of the same generation.
I’ll take my encumbered H.264/QuickTime/whatever any day if it means a sharper, more fluid, and more vivid picture for the same amount of data transferred. If you can’t afford the licenses, fair enough — but if licensing is not the issue, and if you’re choosing Ogg Theora just to keep the freetards happy, please, please stop.
Please look at this comparison ( http://people.xiph.org/~greg/video/ytcompare/comparison.html ) rather than just going on here-say or old news. Theora 1.0 was not quite as good as H.264, but Theora 1.1 will be just as good it seems. Of course by your use of a term like “free-tards” you come off as just another troll, so maybe it’s not even worth bothering trying to talk in terms of logic…
Edited 2009-07-02 19:47 UTC
I’ve heard that one before!
“Version X wasn’t that great, but Version Y will be better!”
Everyone from Linux nerds to Microsoft execs to Apple engineers has used this. Unfortunately, when it comes to something subjective like video quality, such claims rarely pan out — especially when it’s an open-source project.
Also, the results for H.264 from that link are incorrect — http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2009-May/069239.ht…
Disclaimer: I have not reviewed the differences, but you’re leaving out an important factor that may come into play: Decoder complexity. If it takes more CPU power to decode one or the other, that’s another hit against it – in a world quickly moving to low-power devices, the amount of CPU it takes to decode a video stream is relevant as well.
My Atom N270 can barely decode full screen video as it is
Power/datarate consumed is an interesting factor too: depending on the mode of transport (cellular, wifi, wimax), there may be more power savings from having the processor/embedded hardware codec work harder to ensure that you can transfer fewer bits because the per-byte energy of the data transmission turns out to be worse than the energy spent decoding. Also, sometimes data transfer costs become a big factor (some cellular companies are charging thousands of dollars per GB after you exceed your cap.
Notice the recent inclusion of nVidia^aEURTMs Ion with quite a few Atom systems? Desktop GPUs supporting programmable pipelines?
VDPAU / DxVA / XvBA should solve the problems of low power decoding, and highlight certain codecs qualities more (Theora in this case, less overhead), at least on the x86 platform
Ogg Vorbis has been proven, time and time again, to have higher quality per unit of data than other codecs like mp3. Still mp3 is the default and only ‘freaks’ use Ogg vorbis.
Not quite true. Plenty of things use Ogg Vorbis in “embedded” contexts like in-game music, sound effects, etc. It’s just on the consumer-visible side of things that Ogg Vorbis is considered a second-tier option. Even on the consumer side of things, OGG support would be more widespead if the IPod supported it. I have friends who KNOW that OGG sounds better and takes less space, but they can’t use it as their media player doesn’t support it.
Edited 2009-07-02 20:44 UTC
Audio codecs and MP3 debates are yesterday’s news. Today is all about video codecs.
Know why Vorbis didn’t make it? Because iPods and iPhones don’t play it. Whine all you like about unfairness and patented algorithms, but that’s basically what it comes down to.
Also, people who re-encode their MP3s into Vorbis just to be “free” and/or claim better audio quality COMPLETELY miss the point.
In other words, wide support for a codec is more important than pure quality? Uhm, what was your point about Theora vs H.264 again?
On an unrelated note, I think you have a serious attitude problem towards what you call “freetards”. You act as if you have a grudge against them because a horde of “Open Sore Hippies” ate your dog and burned your house. Chill man, nobody here is advocating reencoding their MP3 collection for “freedom”.
That said, you are dismissing the whole patent thing as a non-issue and acting as if only losers would complain about it. Imagine if someone charges licensing fees for HTML so that only big and rich companies can afford making websites. Is everybody who wants to make a website but can’t because of licensing fees “freetards” or “losers”?
Edited 2009-07-03 09:17 UTC
Doesn’t Ogg Vorbis require floating point operations to decode at a reasonable quality vs performance rate? Most ARM processors (usually present in MP3 players) till recently were too slow to handle integer based decoding at a high quality and did not support floating point operations without a software emulator.
Please correct me if I’m wrong.
@memson: Nope. A very good integer decoder for Vorbis has been available for quite some time and even fairly pitiful and underpowered hardware like Sandisk’s Sansa e200 can handle Vorbis decoding VERY well. Heck the thing can even do FLAC when provided with the Rockbox firmware, though battery life is pitiful when doing FLAC as compared to Vorbis or MP3.
Same reason Betamax was a big success over VHS.
Theora has very recently just about caught up. There is only 2db difference now between the development version of Theora encoder, called Thusnelda, and H.264.
Thusnelda still has a bit of tuning to complete, and there is potential that after that tuning Theora could actually be better than H.264.
Please try to keep up, Thom.
Oh, BTW, if you are choosing patent-encumbered H.264/QuickTime/whatever that performs no better than oepn codecs just to keep big corporations happy, please, please stop.
Actually, 2 dB of PSNR is quite a significant difference in image quality measurement. The average gap between H.263 and the reference H.264/AVC implementation is about 3 dB.
Second, it’s just a mathematical metric. It doesn’t take account of our visual system. It’s not uncommon to have images with an higher PSNR that look worse than another with a lower PSNR.
Last, but not the least, one of your own links that you use as reference[1] mention that PSNR is a poor metric to compare different codecs! Don’t you read your own sources?
Like I mentioned in the news about Mozilla Firefox 3.5, H.264 is not a still target. H.264/AVC is fairly complex standard; there are many features that are barely exploited at the moment. Most comparisons between Theora and H.264 are made with the free x264 encoder. There are better encoders out there, like the one from Nero.
The only thing I agree with you is that we should promote open, free standards. For the rest, it’s pure “free” zealotry. Being a “free” advocate is great, being a zealot makes you blind.
[1]: http://web.mit.edu/xiphmont/Public/theora/demo7.html
What is your PROBLEM?
Theora is better already than most of the video that exists today on the web. Most of it is encoded h263 or VP6. Using Theora (decoder) and Thusnelda (encoder), even as it stands now, would be an IMPROVEMENT.
H264 is patent encumbered, and the owners of the patent have announced an intention that will (after 2010) make it prohibitively expensive to adopt for websites that serve a large number of videos (sites such as YouTube).
Better for everyone to adopt Theora now than move from h263 to h264. Its a no-brainer.
Here is an alternative viewpoint that you might want to consider:
http://www.linuxtoday.com/infrastructure/2009070300135OPBZCY
Hear hear.
Edited 2009-07-03 00:48 UTC
My problem is your lack of rigor. You are trying to shove your viewpoints with stuff you don’t understand and keep launching straw mans or red herrings.
I do agree. Yet, just like H.264 and Theora, there is still room for improvement with H.263. It even got features that are not present in Theora (like B-frames).
You keep repeating the same song over and over, yet failed to back up this claim. I would gladly consider an authoritative source.
I always smile when I read about “greed”. Without “greedy” entrepreneurs, engineers and “greedy” opportunities (copyright, patents), chances that we would still be in the stone age of computing. “Greed” is what drives technology. Open software is a great counterpart for keeping the balance, especially when dealing with communications.
So, here’s another alternate viewpoint: believe in what you want, but keep perspective.
The Internet would be better with Theora? That, I don’t disagree, but I don’t expect a change overnight.
Sorry, but it is you who has failed to show any understanding.
You claim to be researching this stuff, yet you didn’t even know anything about the change to the royalties that is going to apply after 2010. You claim that I lack rigour, yet you ignore the links I gave to quotes from the h264 patent holders saying exactly as I claimed, and you were ignorant of, and yet you have the chutzpah to then claim that I somehow needed to provide some backup for what I said?
Pfft.
The “horses mouth” isn’t good enough for you? Video websites scrambling around investigating alternative codecs, and a few of them already going to the trouble of re-encoding their large libraries (another fact you were unaware of), isn’t “backup” enough? What kind of donkey are you, anyway? You call that “research”?
EPIC FAIL.
Edited 2009-07-05 10:51 UTC
Let’s take a snippet from the MPEGLA press release[1] you provided:
It says that Internet broadcast fees won’t exceed OTA broadcast fees during the next renewal term (2011-2016).
It doesn’t say what will be inside the renewal terms in 2011. It doesn’t say what will be the maximum royalty for Internet broadcasts. It doesn’t even say that Internet broadcasts will have royalties. Likely? Sure. Confirmed? Not by that press release.
That is my problem. Even if it’s likely, I don’t want to deal with speculations. Two years is quite a long time in IT; who knows what might happen. With the rise of alternatives, they might grant a royalty exemption for non-commercial usage while questioning the IP status regarding Theora to keep their commercial licensees. Still speculations though, so we will know when they will announce the new terms.
Of course, if you do have sources that deal with the next licensing terms, then please share it to us.
[1]: http://www.mpegla.com/news/n_03-11-17_avc.html
First, you can’t know everything. At least I’m not rampaging the forum with links I might not even have read.
So far, “video websites” are limited to DailyMotion and possibly YouTube. Perhaps there are others, but let’s take these two.
I don’t think Google is motivated by the possible licensing fees. As for DailyMotion, I’ve read their blog entry. It looks like they want to get rid of Flash in favor of HTML5 rather than ditch H.264/VP6 for Theora. They do mention philosophical motivations, but allow me to be skeptical on this, as it’s easy to preach virtue.
Ah, the ad hominem attack. The cornerstone of a bad argumentation.
Since you asked for it, I don’t “research” about H.264. That would be stupid, not to mention useless. I’m researching techniques to improve video compression, targeting but not limited to H.264. I am not an expert on its history, but I am familiar with the algorithms behind video compression. When you spread claims that shows a lack of knowledge (Theora will surpass H.264, the difference between both codecs is “only” 2 dB), then why I should take the rest as gospel?
The sad part is that we both agree that Theora is a good thing; we just don’t seem to agree with the methodology!
I googled “html5 video theora”.
Here are the sites I found, apart from Dailymotion and YouTube, in less than two minutes:
Wikimedia Commons (which hosts over 4,500,000 multimedia files)
http://wikimediafoundation.org/wiki/Spotlight_on_Wikimedia_Commons
The Science Media Network
http://www.sciencemedianetwork.org/HTML_5_video_tag_and_Ogg
Archive.org
http://www.archive.org/details/movies
A list from Xiph.org
http://wiki.xiph.org/index.php/List_of_Theora_videos
The Video Bay
http://pinstack.blogspot.com/2009/06/video-bay-launching-with-html-…
There was another large site mentioned in my original reading on this topic, but I can’t find reference to it now.
Agreed. It is far more likely to be about money, and Dailymotion trying to save some. What is wrong with that, BTW? Especially if it also saves your audience/customers money and/or hassle as well.
Agreed … but you started it … mine was just illl considered retaliation to your original bad argumentation.
Here are some more references to help you:
http://arstechnica.com/open-source/news/2009/07/decoding-the-html-5…
The Ars article gives a fairly good overall summary of this issue.
Under the heading of “The compression efficiency debate” there are some strong points made that are pertinent to what you are apparently trying to research.
http://www.streaminglearningcenter.com/articles/46/1/H264-Royalties…
http://www.streamingmedia.com/article.asp?id=11011&page=1&c=7
Enjoy … or perhaps not.
Edited 2009-07-06 02:55 UTC
The second part is definitely more interesting than the one you have highlighted. If the MPEGLA haven’t decided what they will do, then we are only speculating. For all we know, perhaps my own speculation (royalty-free for non-commercial usage) could turn out the right guess!
Now, I understand why content distributors are worried, especially with the wording of the current licensing terms while needing to make a choice for their future. Therefore, the search for alternatives like Theora is completely justified. Still, it only support my point that we don’t know the next licensing terms right now. This is what I claim from the beginning. Is that clear enough?
Apparently? Well, you just made my day! You don’t know what I am doing exactly, even less who I am, yet you dismiss it completely? Who are you to make such calls? And what’s the deal with these “strong points”? Are you trying to imply that I am wasting my time with H.264? I am quite sure that I mentioned that H.264 is only a target, as I am researching an algorithm. It might give a better or worse compression, hence why the research.
In case you are interested, I would gladly share you my subject and why H.264 in a PM (not because of confidentiality, but rather because we are drifting wayyyyyyy off-topic).
I can understand why some people are strong free (“libre”) advocates, but I don’t understand such zealotry against patents or anything that isn’t “libre”. You are using hundreds of patented design in your daily life, yet I am sure you are enjoying these contraptions…
Edited 2009-07-06 06:00 UTC
Patents are not supposed to apply to mathematics. Results of scientific and mathematical research (such as what you yourself apparently are involved in, and I say apparently only to indicate that I cannot verify what you are involved in, and not as any intended or implied disbelief) are supposed to be community knowledge.
Patents are for inventions. Applied knowledge. A particular “method” of doing such-and-such a task.
Ergo, no patent should apply to a codec. You cannot get anything more like pure mathematics than that. It is simply a mathematical transform from one stream of data to another using an algorithm.
So when a company wants to use a specific piece of mathematics to extract a royalty out of the community, especially when there is copious prior art on the “methods” for employing that mathematics (i.e. lots of prior art on the general idea of codecs themselves) it is more than just prudent, it is nigh-on imperative, to oppose that company being able to extract any monetary reward from gaming the patent system.
This is not “zealotry” of any kind, it is simply plain protecting oneself and other community members from rip-off merchants.
Hey Thom, here is an alternative viewpoint that you might like to consider:
http://www.linuxtoday.com/infrastructure/2009070300135OPBZCY
Cheers.
Because it’s the only one that everyone is capable of standardizing on. It’s not just “freetards” that would be boned.
I agree with you as far as personal use is concerned, but if it’s something everyone is supposed to be able to use..
So why can’t someone make a Theora codec plugin for Quicktime, which will allow Safari (which relies on Quicktime) to support Ogg Theora as well?
http://xiph.org/quicktime/
There is a ogg theora quicktime plugin already: http://www.xiph.org/quicktime/
Apple dislikes OGG Theora mostly because it can’t be implemented on the iPhone. Using the CPU to decode it will be too slow and the h264 decoding hardware is no good for Theora video.
Also it is unlikely that Google will encode the youtube videos in Theora format because the quality is not that good compared to H264. Sadly Theora is actually based on an old tech (VP3) and there’s not much juice left that can be squeezed from it.
Maybe in a few years, when the Dirac codec will be mature enough and the handsets will have enough power to decode it we will have a standard…
Edited 2009-07-02 19:40 UTC
I think you’ve got the whole thing figured out about why Apple can’t stand to use Theora. If I hadn’t already posted in this thread, I’d have modded you up “insightful”.
Theora 1.1 (codename: thusnelda) should be just as good as Mpeg4/H.264 once it is complete.
The real problem is in the hardware like you said. Currently there are no off-the-shelf chips that support Ogg codecs, requiring the CPU and/or GPU to handle it all by themselves. Unfortunately we’re in that whole chicken and the egg scenario on that deptartment. No one wants to create hardware encoder/decoders until there’s a larger market, market won’t expand until there is hardware :/
Their is however design information in VHDL.
This is just not true.
The Theora bitstream format is frozen. Any Theora decoder made since 2004 will be able to decode and render Theora videos made aven now with later technology encoders.
And exactly how is the situation with encoders for Theora?
Well, Theora encoder (in the form of the still-experimental Thusnelda encoder) have surpassed h.263 and almost caught up to H.264 (only 2db in it).
http://web.mit.edu/xiphmont/Public/theora/demo7.html
Now H.263 is used for the majority of video currently on the web (read … on Youtube). Theora would be better than that. Google’s argument is actually bogus.
H.264 is clearly superior at this moment. No question about it. But why doesn’t Google pour the money into the Ogg project they would have otherwise spent on paying royalties? This would definitely improve their reputation and it doesn’t cost them a lot of money. This might close the quality gap at least a bit, depending on the current state of Ogg Theora.
But I don’t really think this debate matters that much. Whatever Google (youtube) will choose will be supported by all browsers except maybe in Debian .
As for audio: I’m a huge proponent of HE-AAC.
http://tech.ebu.ch/docs/tech/tech3324.pdf:
Note that this is about multichannel 5.1 audio, so when comparing to stereo divide by 3! So only 160/3=53kbit/s would be required to make sure a stereo audio file encoded as HE-AAC would still sound “Excellent”.
I believe AAC only requires a royalty if you develop a binary codec, ie a encoder and a decoder. Distribution of content is entirely royalty free. Everybody wins!
AAC still has patents on it, so yes there are royalties, and if you’re using a special version like HE-AAC there are even more licensing issues.
Ogg Vorbis supports 5.1 audio too. In various listening tests some find one better, while others find the other best…making it negligible as to which is superior on sound fidelity.
You are wrong:
http://www.vialicensing.com/Licensing/AAC_FAQ.cfm?faq=1#1
HE-AAC is also included: http://www.vialicensing.com/Licensing/AAC_FAQ.cfm?faq=3#3
Btw not that I would mind if Vorbis became part of the specs… Au contraire. I just think AAC would be a better choice.
Edited 2009-07-02 20:15 UTC
So you’d be OK with an HTML standard that can’t be implemented (legally) in its entirety by an open source project? If HTML5 were to specify MPEG family codecs, that’s exactly what would happen. And that’s why the W3C decided to specify NO preferred codecs if they couldn’t get the Ogg family approved. Unlike ISO, which is completely beholden to corporate interests and has no problem allowing standards that are licensed under so-called RAND terms (nominally Reasonable and Non-discriminatory, but in actuality Unreasonable and VERY Discriminatory), the W3C believes in OPEN standards: anyone can implement them using whatever source model they choose. Hats off to the W3C for not compromising its principles!
Are you saying that MPEG standards are acceptable? Because, well, AAC is defined by both MPEG-2 part 7 and MPEG-4 part 3 and by ISO.
http://en.wikipedia.org/wiki/MPEG-4_Part_3
http://en.wikipedia.org/wiki/MPEG-2#ISO.2FIEC_13818
Besides it is possible to do everything royalty-free. FAAC en FAAD (open source encoder and decoder) didn’t pay a single ^a‘not to anyone.
@LB06: MPEG standards aren’t acceptable at all. While people in places like the EU where software patents don’t really exist can legally create and use software based on MPEG standards without paying any royalties, those in place like the US and Japan, where software patents DO exist have to pay royalties and deal with patents to legally implement those standards. The MPEG is a working group of the ISO. The ISO allows patented elements to be included in their standards and DOESN’T require that companies grant royalty or restriction-free licenses. A few working groups (such as JPEG folks) go above and beyond and actually create OPEN standards, but many others, like the MPEG team, essentially operate as profit generation machines for large multinational companies by working to cram in as many patent-encumbered technologies into standards as possible, thus making the implementation of those standards a sure money-maker for the various companies on the committee in the first place.
Raise your hand for all those hardware vendors supporting Ogg Vorbis stuff. Sorry but software solutions won’t cut it against hardware accelerated ones.
I agree. For this reason I’ve bought Trekstor’s mediaplayers Vibez (plays Ogg Vorbis and FLAC) and MovieStation (plays Ogg Vorbis).
And written them a letter praising the support of Ogg Vorbis
http://wiki.xiph.org/VorbisHardware
You just showed me a list of either niche products that may or may not work with Ogg Vorbis [hence not actually tested by Xiph.org] or a list of obscure vendors that don’t address the general masses.
The Theora list is an absolute joke.
Show me the GPUs from NVidia, ATi, Intel that handle Ogg Vorbis codecs for video and the audio chipsets that manage the sound to accelerate it.
Neither RIMM nor Apple will ever support Ogg on their smartphones.
Android isn’t taking over the market for smartphones.
Palm Pre? Another small market when it comes to smartphones.
The Audio Players listed competing with the iPod/iPod Touch are fractions of a percent in total.
Hell, the entire list of market share is puny and/or specialized.
What? Nothing from BOSE? Nothing from SONY? Pioneer? Panasonic? Philips?
It’s DOA.
That maybe so … but nevertheless it is NOT true to say that no hardware chips exist. They do exist, and they do work.
Your rant is not that there is no hardware support … your rant is actually that hardware product builders and aggregators, for whatever reason, have largely chosen not to include the support that does in fact exist at the component level.
It would actually be in mosts people’s interest if they had. One wonders exactly why they didn’t …
I think you also need:
http://wiki.xiph.org/index.php/Theora_Hardware
Which is pitiful.
You know what is really stupid is, it almost seems like now video-tag doesn’t have any required codecs, audio-tag won’t either. Not even wav or FLAC.
This is really sad, but discussion is still going kind of. So who knows.
Xvid FTW!!
lol
soooo….. h.263 then….
You have to understand the difference between a software product and a format. Xvid is not a video format, Xvid is a software library that encodes and decodes MPEG-4 Part 2 (Advanced Simple Profile) video. It’s just one of many MPEG-4 ASP implementations (codecs). MPEG-4 Part 2 is covered by patents just like H.264 (which belongs to the MPEG-4 family, too), and it’s an outdated, inferior technology. So there is really no point in using it.
…VC-1
MS’s created, and standardized, WMV implimentation.
http://en.wikipedia.org/wiki/VC-1
Though I believe MS is working on something new with better compression to quality ratio since there is comming a bigger focus on small mobile devices with streaming media. more on that will come later…
Possible future codec aside, it’s likely MS will support both VC-1 and H.264 at a minimum. They currently implement both across a number of products (Zune, Xbox, Media Center Extenders, Windows 7, Windows Mobile, Silverlight) and are part of the patent pool for both formats.
I assume firefox has some sort of plugin infrastructure to allow new codecs to be added. A few 30sec googles haven’t turned anything up. Anyone know how I can do this? I use firefox on Linux, OS X and XP/Vista.
Granted, “out of the box” support is great. But adding a codec plugin (hopefully from a choice suppliers) is easy. I really hope firefox was designed for this.
Thanks.
Have the tag support multiple source attributes and the browser can simple use the source that has the video codec it supports. Seriously, h.264, vc1 and Theora are going to be the most popular. If those are supplied by the site creator, then the problem is solved.
This is used all the time for image tags with the alternate attribute.
Fallback in browser support is available:
https://developer.mozilla.org/en/Using_audio_and_video_in_Firefox
Rather I would ask “How may different versions of files content providers bother to encode?”
how many people does the content owner want to view his/her stuff?
Lets get the tag working now and let the browser vendors worry about codec licensing later. I guarantee, if they offer Ogg (supported in FF) and VC-1(Supported in IE I am sure) and h.264(supported in webkit) They will cover everyone that needs to worry about it. Eventually the codecs will get worked out and everyone will allow free decoding capabilities to the other vendors. What is really at issue however is as video improves, will browser vendors implement the new standards? jpeg2000 anyone? The best thing to do is provide codec plugins for the browsers. The video tag is really there to make it easy for the site developer.
Edited 2009-07-03 02:56 UTC
If you run a site that hosts/publishes video content and that’s your primary goal (having your content viewable by as many users as possible, without the need for them to install additional software), then there is already a technology which meets that requirement: Flash. While there are certainly issues with Flash video, they are (IMO) not severe enough to provide a compelling reason to ditch the technology altogether – at least not when the alternative solution requires 3 separate versions of each video instead of one (along with the extra time to create the additional versions of each video, both in terms of human effort and processing power, and the 3x increase in diskspace needed to store the files, etc).
The solution you describe is not much different from what we had in the “bad old days” – except back then you had to provide Quicktime, WMV, and RealMedia versions of each video instead of OGG, h.264 and VC1. And that option effectively lost out to Flash – largely because (again, IMO) Flash support was, and still is, more ubiquitous than support for any of the video formats being pushed as a standard for web video.
So if Flash won out due to its ubiquity, then it stands to reason that the only way to displace Flash is with a technology that’s more ubiquitous than Flash (and therefore more convenient for content publishers). The HTML5 video tag could provide that – but not without a video format that’s consistently supported by all of the major browsers.
Of course, this all assumes that the use of Flash video is an issue that needs to be “fixed” somehow (a premise that I don’t entirely agree with).
if you are the maintainer of the site, being able to maintain one document as opposed to more than one, and use one tag as opposed to an application, encoding 3 versions of your video is easier.
I’m afraid the point of your comment has escaped me. Why would using the video tag (along with 3 versions of each video) require fewer documents or tags? If anything, wouldn’t it be the reverse?
Ooh, I can^aEURTMt wait for 3 codecs used on the Web! Oh wait, no, 4.
Ah, 5, shit!
6?!?
etc
^aEUR| a situation you cannot contain once you open it up.
What right does W3C have to force any specific video format on everyone else? If they choose something today, what will they do when their chosen format becomes obsolete?
Video, animated pictures and plain pictures should all be treated exactly the same by W3C – some sort of “content” tag, with a URL to a file in the format preferred by the web site author, and optionally more URLs to more files in alternative formats. For e.g. “<content = http://www.foo.org/bar/a.png, http://www.foo.org/bar/a.jpg, http://www.foo.org/bar/a.gif>Alternative text</content>”.
The other thing I’m wondering; if it took up to 5 years for the major browsers to support HTML 4 properly (e.g. including passing the ACID test), then has the HTML format become an over-complicated bloated mess that needs to be discarded?
What right does MPEGLA?
Improvements are driven by the encoder. The decoders are capable of still decoding the improved files. Hence the Ogg Theora decoders being delivered in Firfox today will be able to show Theora video well into the future.
http://www.xiph.org/quicktime/download.html
One simple download.
Please pass this on.
For all the wild claims about it’s existing ‘quality’ or how some future version will bring it up to the quality of h.264 – the fact is it remains in ‘catchup’ mode and is likely to remain so since there is no revenue to pay anyone to actually work on it.
There, I said it… Funding by paypal donations is a world away from a serious company.
The ONLY reason it is touted so highly is the same rampant FLOSS Fanboy bull that makes dozens of other tinker-toy codebases be exaggerated in capabilities – It’s just more of the “fight the power – down with the evil corporations” dirty hippy nonsense that should make anyone with a brain jerk into Eric Cartman mode – because all that jamming to a crunchy groove and how it’s going to put the Eichmann’s in their place is not going to change a damned thing. Do us all a favor, and go back to your drum circles.
I keep seeing phrases like ‘quality per pixel’ – Bull. What I think it meant, or should be meant is quality per BIT. Even if you take the lack of hardware support out of the equation you are still stuck with an encoder that even at the most up to date version (Thusnelda?) has horrible artifacting at a whopping 360K/sec (that’s bytes not bits – 2900kb/sec) for a 640×480 video that makes 50% jpeg encoding look GOOD.
Take a look at the x264 and then the thusnelda on this page:
http://saintdevelopment.com/media/
Free/free (both beer and freedom) rarely produces anything of equal quality to commercial products unless the people working on it for free use it in their work and get paid to work on it. See Linux, Apache, etc… or if the people who use it have the skill to fix things they don’t like – see firefox.
Unless they miraculously get people who want to encode videos and get paid to work with video AND have the skills to contribute to the codebase, it’s going nowhere as free/free. The other alternative is to put a legitimate and practical business model behind it, which would piss off the zealots no end.
Sometimes things are actually WORTH money, it’s why people generally do WORK is so they can EARN money… Naive pipedream bull to the contrary is NOT a viable survival plan.
Oh, and YES, I realize Chris ‘works’ for Red Hat, allegedly on Ogg…
Edited 2009-07-03 01:48 UTC
So are you advocating a world in which only people with enough money have the right to publish videos that everybody can easily view?
Player like firefox or opera or any decent alternative browser, can’t seriously afford the licence for h264.
That leave the big wallet guy MS,Google,Apple in the race of mordern browser.
Hell I remember that neither IRC , Newsgroup gopher or www (html3/4 or xthtml ) or reading my mail do not require expensive software to be used.
I seriously don’t care that much about video integration into the browser ( I hate putting all my egg in the same basket ).
And Apple is mostly protecting its investisment with mpg4 h264 as there is already a port of vlc on the iphone which, I guess, can support ogg format.
Of course they must choose ogg format because the web must based on open standards and ogg is open source format
Just make a new version of the ACID test (ACID 4) and have it include an Ogg Theora video. Apple will immediately release a new version of Safari that supports Theora, just so they can say that they’re the first to pass ACID 4.
The article states “Apple has said that they will not implement Ogg Theora support in QuickTime (what Safari uses), citing a lack of hardware support and an uncertain patent situation.”
The linked source, however, does not state that. So, where does this claim come from?
why can’t they support more than one video format? they do it with images. yeah using multiple formats can mess up the portable web. but it wont mess up computers. with everyone making iphone websites though, most things are going to use apple’s format.
The idea was to add a requirement about what should be codec should atleast be implemented. The video-, audio- and image-tags all support multiple formats, actually if things go as they look like they might now be, non has any requirements about what formats you need to support if you want to support a spec. Having a baseline would be really useful though, even if it was FLAC or wav for audio.