Talk:Theora

From Wikipedia, the free encyclopedia

To-do list for Theora: edit · history · watch · refresh

Here are some tasks you can do:
  • Requests: Expand the article. If possible make it more clear for non-techies.
  • Cleanup: Clean up for a GA rate.
Priority 4  

Contents

[edit] Helix links

Here "RealNetworks Helix player" contains two links to the same place (becuase of redirection) - should this be fixed?

[edit] VP3 license

I removed the above from the article. The Theora sourcecode isn't based on that license, so it would belong somewhere else. Kjoonlee 04:01, 2004 Dec 15 (UTC)

That is no longer true. The VP3 article has been merged into Theora. Rcooley 11:38, 23 October 2007 (UTC)

Yes, it was, but what's that got to do with anything? Kjoonlee point remains valid.--Ivo Emanuel Gonçalves talk / contribs (join WP:PT) 14:03, 13 November 2007 (UTC)

I've added the VP3.2 license back into the article, as a reference under VP3. Rcooley (talk) 08:33, 11 February 2008 (UTC)

[edit] Use on Wikipedia?

The Theora and Vorbis royalty-free, open source video and audio codecs could help to make it possible for Wikipedia to host narrated video clips for those articles that explain processes or mechanical devices which are difficult to comprehend using words and pictures alone.

If anyone can point me to some examples of Ogg Vorbis/Theora being used here on Wikipedia, I would be most grateful, as I would like to contribute video and audio clips if it is possible to do so. — DV 07:52, 22 Dec 2004 (UTC)

see Wikipedia:Sound and m:Video policy --Hhielscher 20:27, 26 Mar 2005 (UTC)
Apollo 15 is a nice article with links to ogg/Theora movies.

[edit] Lossy or Lossless ?

It doesn't say if it is lossy or lossless compression, any talk of compression should say that off the bat. Does anyone know, this is outside my knowledge base but I'll look around. --ShaunMacPherson 22:20, 2 Jun 2005 (UTC)

Since almost all video codecs are lossy most articles in Category:Video_codecs don't talk about it, except the lossless SheerVideo and Huffyuv. Unfortunatelly the same states for Category:Audio_codecs. Theora is lossy as well. --Hhielscher 23:52, 2 Jun 2005 (UTC)

[edit] quality comparison?

Is there any information out there comparing the quality of Theora with other codecs? I know this can be hard, since people get subjective, but it would be nice if the article mentioned which codecs theora has a better quality/compression ratio than, those it is very similar too, and those (if any) that might have better quality/compression ratios than it. Jet

I've updated the Theora article to include a citation to Doom9's 2002 Codec Comparison, 2005 Comparison Qualifying Round, and Monty's own write-up on the limitations and poor quality of Theora. Rcooley (talk) 23:17, 11 December 2007 (UTC)

Is a 2 year old comparison still a valid source for describing the state of Theora today? Especially since version 1.0 Beta 2 was only released in Oct 07. Also, "Monty's own write-up" is actually an archived paper discussing improvements being considered on the way to the 1.0 release. Is it still relevant (especially under the VP3 description header)? --Dl036 (talk) 23:16, 15 December 2007 (UTC)

Until someone performs some new quality comparison references, that disagree with the current ones, it absolutely should stay, no matter how much time has passed since it was written. However, I wouldn't strongly object to the statement being moved out of the VP3 section, but someone will have to make an appropriate place for it. As the statement applies to both VP3 and Theora it will necessarily seem out of place under either a VP3-only or a Theora-only section, and those are the only two options right now. Until someone steps in to clean-up the entire article, that will probably remain necessary. Rcooley (talk) 04:03, 18 December 2007 (UTC)
What, because no one has peformed a similar test in three years, we should accept these results as currently accurate? That makes no sense. Maybe you could say that VP3 and early versions of Theora were comparable to MPEG-1, if that were true (as I said, I don't know), though as I said below, neither of the articles you cite make that claim. As it is now, however, I believe the statement is misleading. --Dl036 (talk) 01:15, 21 December 2007 (UTC)
Personally, I do not feel that comparison is a valid source, and Theora is not quality-wise similar to MPEG-1. I don't think it ever was unless one's using the very very earlier versions of the official encoder. However, due to my relation with Xiph I cannot modify that statement myself or I would appear biased. It would be great if someone would grab an easy to use encoder like ffmpeg2theora and try it to confirm that those claims are unlikely to be correct.
Actually, on checking those references again, neither Monty nor Doom9 seem to ever refer Theora as akin to MPEG-1 in terms of quality. They only say it's not as great as other current video formats. Of note that Monty has also said (on another occasion) that Theora can be improved to the point of being near or superior in quality to H.264/AVC.--Ivo Emanuel Gonçalves talk / contribs (join WP:PT) 00:29, 16 December 2007 (UTC)
Someone "grab[ing] and easy to use encoder" to check "those claims" would be "original research" and never allowed here. Somewhat besides the point, but I should also mention that while there is only one VP3 and one Theora codec, there are many MPEG-1 codecs, so perhaps your opinion is based on the use of a poor one, or possibly a less than scientific comparison, using different settings for each one. I personally really hope Theora improves to become competitive with MPEG-4 (either ASP or AVC), but that absolutely isn't the current situation, and frankly, with the approx. 5 years it took to get to the current state, I don't expect to see substantial changes anytime soon. Rcooley (talk) 04:03, 18 December 2007 (UTC)
I have never used VP3, so I can not comment on whether the description of it being similar to MPEG-1 is fair. Anyone know if it would be more accurate to simply remove Theora from that particular reference (in which case the citations are not relevant), or if the entire last sentence under the VP3 section should be removed in order to correct this? --Dl036 (talk) 18:55, 16 December 2007 (UTC)
The reference in question repeatedly states the quality issues with Theora are because of code and features directly inherited from VP3, so the citation is pretty clearly relevant to VP3. The 2002 Doom9 quality comparison also directly addresses VP3, but it's only available online today in a RAR archive, and no one has yet suggested an alternative method for citing it, but in any case, it is a very real and verifiable source (that everyone selectively chooses to ignore). Rcooley (talk) 04:03, 18 December 2007 (UTC)
I am not selectively ignoring anything. I read through both the articles you have cited once again and neither of them suggests that Theora is comparable to MPEG-1. Further, both documents are three years old and are speaking about earlier alpha versions of Theora. Your reference may indeed be applicable to VP3 (as I said, I've never used it), but it's not my experience with the current state of Theora and as such I believe it is misleading. --Dl036 (talk) 20:17, 19 December 2007 (UTC)
I guess the only way to please both sides of the issue would be to state VP3 may be similar to MPEG-1 in quality (if a good reference for this exists), while Theora surpasses VP3 in quality (reference exists, but it would come from Xiph and not a third party). This issue is complex...--Ivo Emanuel Gonçalves talk / contribs (join WP:PT) 18:22, 19 December 2007 (UTC)

1) Personal opinions are not authoritative and verifiable sources, and so irrelevant here on Wikipedia. -- 2) It's pure speculation to assume that just because 3 years have passed, that substantial changes have occurred... It's possible, but the opposite is equally possible. Baring any citations to support such claims, there is no reason to assume the existing citation are incorrect in any way. -- 3) I have changed "MPEG-1" to "h.261" (which makes Theora's quality sound even worse, IMHO) and added yet another citation to substantiate that statement. Hopefully that will resolve this insignificant sticking point. -- If you have a source that contradicts any of the 4 sources I've now included, PLEASE mention it. Otherwise, the statements about quality should (and will) stay as-is. I'm not going to waste more of my time on this talk page, arguing with everyone who is trying to change facts to get it to agree with their dogma. This is not the Xiph.org Wiki. It's citations, or nothing at all, folks. So far, everyone else has provided exactly that... nothing. Rcooley (talk) 23:59, 19 December 2007 (UTC)

The citation you have posted that compares Theora to h.261 clearly states it is the author's subjective opinion, and therefore is just as unacceptable as anyone else's speculation. I am not trying to make this the Xiph Wiki, nor do I have any vested interested in this matter. Theora is one of a number of codecs I work with regularly, and I simply know that your likening it to MPEG-1 or h.261 is wrong (or at least out of date). And the citations you use are, again in my opinion, either too dated or don't really support your statements. If I had an agenda I would have continued re-editing the article before reaching a consensus on this discussion page, as you have done, but my only concern is that people reading your description will be mislead. --Dl036 (talk) 15:31, 20 December 2007 (UTC)
Quality comparisons between lossy codecs are inherently based on some personal opinion, as eyes (and ears) vary. He also may have wanted to state that it isn't Nokia's official position. In any case, he specifically said his conclusion was based on tests, and he did NOT state or imply it was speculation. Finally, using YOUR OWN opinion to decide is directly in conflict with Wikipedia's policy on Citations and Original Research. A citation with a "subjective opinion" is not. Rcooley (talk) 19:35, 20 December 2007 (UTC)
Rcooley. The last edit wasn't made neither by Dl036 or me, and it was a bad idea not to read that anonymous editor's rationale, which was that neither Nokia's sensasionalist claim or Doom9's 2002 review in RAR are good references. Also, Monty's reference which you also want to add explicity refute your own claims, so why do you keep adding those bad references back? I don't get it. You keep saying we need to add citations to support an argument, but then you go ahead and add bad citations, and when two different people decide to remove the argument altogether — not refute it, remove it as to avoid conflict — you keep reverting their edits. Seriously, just cool off and listen to the other editors. Wikipedia is run by common sense and collaboration, and I think you would do well in trying them both.--Ivo Emanuel Gonçalves talk / contribs (join WP:PT) 23:47, 20 December 2007 (UTC)
Apologies to Dl036. He seems to be logged-out half the time he edits wikipedia, so when I saw his talk post shortly thereafter, I just assumed it was his IP address and didn't really check. (I'll remove the statement suggesting he was responsible.) I did read the rationale for the removal and it's simply nonsense from beginning to end... The Nokia paper says h.261 is about as good as Theora, and Monty says Theora isn't "worse than h.261". No conflict there. And Monty's statement was also just a recent flippant add-on, not otherwise a part of his paper. I can't see anything wrong with the Nokia paper, and you haven't given any reasons either. I also haven't seen any reason RAR archives of webpages don't make an appropriate source. If it can be found online, I can't see a reason not to cite it. I could always book/publication-cite it, but I don't see any reason to do so. The HTML pages loads now, and readers can chose to download it, or not. If someone could justify wanting to remove a single citation, rather than insisting on obliterating the statement (as has happened twice now) I likely wouldn't object. The other sources are adequate, but it's always better to give extra info. "Avoiding conflict" with dogma is not a justification for removing very well-supported facts... Rcooley (talk) 00:45, 21 December 2007 (UTC)
Looking at the citations: Wenger, Stephan (2007), Web Architecture and Codec Considerations for Audio-Visual Services, pp. 3, <http://www.w3.org/2007/08/video/positions/Nokia.pdf>. Retrieved on 2007-12-19 seems to be the author's opinion with nothing of substance (stills, clips, graphs) backing it. And Doom9 (2002), MPEG-4 Codec shoot-out 2002 - 1st installment, <http://www.doom9.org/codec-comparisons.htm>. Retrieved on 2007-12-19 is far too old to be of use. I think that to compare Theora in quality specifically to H.261 We really need something more substantial than Wenger's personal opinion. --XanderJ (talk) 10:46, 25 January 2008 (UTC)
Wenger is an expert, talking about the subject at hand, and therefore a verifiable source. You can't get any better than that. Wikipedia is not a collection of side-by-side comparisons so you can "judge for yourself." Furthermore, his conclusions are in-line and fully supported by every single other source on the subject. The "2002 Codec shoot-out" is about VP3, not Theora. VP3 hasn't been under active development, so the age should not be an issue at all. I must also stress what I've said before... even if the source was about something still developing, age should not be a disqualifying trait, unless multiple, newer sources directly refute it. Improvement over time may be a possibility, but it is certainly not a given. Rcooley (talk) 19:53, 28 January 2008 (UTC)
If a source is only about VP3, then it should not be used to describe "VP3's and Theora's performance," such a a claim is misleading. --XanderJ (talk) 09:09, 1 February 2008 (UTC)
Wenger works on MPEG standards and he wrote that position paper for Nokia (an MPEG company). His claim even was disputed in mainstream media. http://www.heise.de/newsticker/meldung/100380/ - "It's incomprehensible how Wenger, who dealt with audio- and video-coding for years, arrives at the conclusion that Ogg Theora and H.261 are on the same quality level." (translated from German "Wie Wenger, der sich selbst jahrelang mit der Kodierung von Audio- und Videosignalen beschäftigt hat, zu der Ansicht gelangt, H.261 und Ogg Theora bewegten sich qualitativ auf dem gleichen Niveau, bleibt allerdings schleierhaft."). This article is written by Dr. Volker Zota, who's working at the Magazin für Computertechnik (c't), Germany's most renowned computer magazine. --Maikmerten (talk) 11:55, 9 February 2008 (UTC)

[edit] patent-free ?

Are there any proof ? I was said that near everything in video is patented (even the arithmetic sum). Moreover derf experimental theora codec lists some patents : http://svn.xiph.org/experimental/derf/theora-exp/doc/patents.txt

That link doesn't mean anything. Um, got any proof that Theora infringes on someone else's patents? Besides On2's, that is.
First of all nobody can guarantee for any product that it does not infringe some patents, which is a major problem for the whole patent system. However, as all basic coding methods used in Theora are well known since the 1980ies (meaning any patents would have expired by now) or are subject to the VP3 patents for which a license was donated up till now noone was able to come up with a patent infringed by Theora. And even if e.g. an encoder was found to use patented compression methods this doesn't mean that other encoders, players or the bitstream specification itself are having this problem, too. --Maikmerten (talk) 13:35, 27 January 2008 (UTC)

[edit] Non-humans?

Ok, this is weird, but did they give licenses to other sentient beings, or just humans? You never know what will come up tomorow, you know.....

I don't know. However I have made the claim more neutral as below:
While VP3 is patented technology, On2 has irrevocably given royalty-free license of the VP3 patents to everyone, letting anyone use Theora and other VP3-derived codecs for any purpose.
However I still feel it might need to be reworked. If On2 did explicitly say they were releasing it to all humanity in their license agreement and therefore technically if you weren't human it wouldn't be release to you, you're welcome to change it back of course Nil Einne 13:03, 18 June 2006 (UTC)

[edit] Remove history sections?

I find the current 'Development timeline' and 'History' sections too verbose and not informative. Anyone interested in the changelogs would likely be looking elsewhere, not on Wikipedia, and for everyone else it fails to be useful and just takes up space. Does anyone agree with me that they're best removed entirely? -- intgr 22:02, 3 August 2006 (UTC)

This information is really unnecessary and do not contribute anything to the article. 130.233.243.229 12:28, 10 October 2006 (UTC)
Done. -- intgr 15:39, 10 October 2006 (UTC)

[edit] Tarkin

Hi-

Should we include a link to Tarkin, as it's another codec that the Xiph.org Foundation plans to implement sometime in the future?

Stilroc 02:53, 26 September 2006 (UTC)

This is a late reply, but alright. Xiph does not intend to return to Tarkin any time soon. It was pretty much dropped for Theora. There's a link to Tarkin in the Xiph Projects template, though, so that should be enough.--Saoshyant talk / contribs (please join WP:Portugal or WP:SPOKEN) 13:00, 22 February 2007 (UTC)

[edit] Chroma bleeding or colour bleeding?

There is no wiki page for "chroma bleeding". I was going to submit some information after researching the topic, but found little. Is this artifact more commonly referred to as "colour bleeding"? And should the the link therefore be updated to reflect this? Jwitch 13:22, 28 December 2006 (UTC)

Color bleeding. But I removed the text. Theora used subsampled color, like all the other video codecs anyone cares about today, and also like all the other codecs it operates in YUV space. Subsampled color will always have some bleeding potential, but the use of a color space which is not perceptually linear makes it more likely for the bleeding to be noticed. Theora could have been designed to use a space with Y gamma corrected to perceptual linearity, but this would have required an additional set of per pixel multiplications on decode side.. removing the advantage of CPU saving hardware YUV to RGB conversion which is widely available and used. We might want to have some text on subsampled color, but to single out Theora on this matter isn't fair or accurate. --Gmaxwell 16:55, 28 December 2006 (UTC)

[edit] Adoption

The video game Heroes of Might and Magic V uses ogg for its cinematics, complete with Theora for video and Vorbis for audio. Should we mention it somewhere? - Sikon 12:38, 27 February 2007 (UTC)

Wikipedia may not be the best place for this. Xiph's official Wiki provides pages to list players, software and games that use their formats. I've just created the page Games that use Theora using your information. If you'd like to add more information, I believe this is the place.--Saoshyant talk / contribs (please join WP:Portugal or WP:SPOKEN) 13:17, 27 February 2007 (UTC)

[edit] x264 in Matroska

With Mediacoder 0.6.0, I can convert to x264 (aka FFH.264) video /w Matroska container, but sound will only work with mp3 (still LAME). Point is everything here is open source so i may use that, because the quality at 350 kbps a lot better then Theroa. Renegadeviking

"Open Source" doesn't mean "free". x264 is encoding H.264, for which patent fees are collected. Lame encodes to MP3, for which patent fees are collected as well. --Maikmerten (talk) 12:19, 9 February 2008 (UTC)

[edit] Opera 10 Browser to support Theora

Interesting video: http://coolastory.blogspot.com/2007/03/sv-web-builders-event-world-premier-of.html

Right at the end, the demo of Opera 10, and the mention of Theora.. —The preceding unsigned comment was added by 81.174.171.21 (talk) 17:19, 13 April 2007 (UTC).

Opera released an experimental Windows version with Theora support, which may be found in here. It's recommended that you read the instructions at "A call for video on the web" if you plan to try it.--Saoshyant talk / contribs (please join WP:Portugal or WP:SPOKEN) 17:52, 20 April 2007 (UTC)