Guest post by Chris Montgomery, otherwise known as Monty, is the founder of Xiph.org foundation and creator of the Ogg container format. He has been sponsored by Red Hat for several years to improve the codec quality of Theora and the next generation version, called Thusnelda, is already proving to be better than H264 as bitrate increases. Monty has posted some test results demonstrating the improvements. Chris Blizzard from Mozilla Foundation has some updates as well.
The German news site heise reports that there has been an error during measuring h.264 and therefore its result were worse than expected. Apparently h.264 still produces better images than Theora but the difference between the two codecs is now far less visible.
http://www.heise.de/newsticker/Freier-Videocodec-Ogg-Theora-ist-H-2… (German)
http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2009-May/069239.ht…
If Theora is close to h264 in terms of quality per kilobit, that is very good.
But I’ll continue to use h264. Strangely, it seems to be better supported in Linux multimedia programs; for instance, some DVD authoring programs will accept h264 as input, but not Ogg Theora! And some programs, like Kdenlive, will output really low-quality Vorbis audio in Theora videos even when you tell it to use a high bitrate and sample rate for the audio.
File bug reports and help improve those programs. Sticking with patent encumbered solutions is not going to help in the longer run.
Well, to be fair, using non-patent-laden solutions probably won’t help much, either.
I disagree with that. It is unfortunate that organizations like Nokia voted down Ogg from HTML5 but I believe that major sites like Wikipedia adopting Ogg as a standard and browsers like Firefox and Opera getting the ability to play ogg out of the box without any plugins along with the rapid quality improvements in the theora codec quality will have a profound impact on the open web. Not to forget BBC and the dirac codec.
Theora will be like VHS; as long as it is good enough for the vast majority of end users and is low cost for content producers and implementers (programmers) – it’ll win over content on the internet.
What I’d love to see, however, is for Google, Adobe (add support inside Flash) and Apple (add official support in Quicktime) get behind it. If that happened; it would whittle down the selection between Microsoft and their expensive CODEC or an open source royalty free video and audio technology which reduce the costs for these said companies.
Actually it will more like Betamax.
Outside the OSS world no one cares about Theora.
That could very well change if its quality became on par with more modern codecs like H.264, or even close. Currently it is overlooked because, let’s face it, its quality is just bad compared to modern codecs, and even some older codecs such as XviD/DivX leave Theora 1.0 in the dust with ease. Were the quality to improve, and to do so in a reasonable amount of time, it may very well start coming into its own outside of the foss world. Mind you, if they take five years or so to officially release this new branch of Theora… they’ll just be left in the dust again.
Personally, I’m not betting on Theora improving in the needed time frame… but I would bet on Dirac becoming popular outside of the foss world in the not too distant future.
I would be quite interested to see Theora in browsers if/when video quality becomes comparable to H.264.
Video playback in a browser requires that you either build some plugin (and nobody likes to install a plugin just to play back video on a website) or that you use flash. Flash is kinda nice, and works mostly, but unfortunately it’s H.264 decoding performance sucks on Linux and Mac.
In my perfect world all major browser creators would buy a blanket H.264 license, so all browsers would be capable of natively playing back H.264 without using any plugins (preferably encapsulated in a transport stream, so it works nicely with live video too).
Sure, flash is nice… if you ignore the fact that it’s a bloated, resource-hogging rat’s nest of instability.
The theora codec, like all codecs, has two components, namely the encoder and the decoder. What one uses in a browser is the decoder. The Theora decoder has been released now at version 1 for some time.
The Theora encoder at version 1 was a pretty poor performer, unfortunately. That impacts the use of Theora in browsers only because you couldn’t encode a decent bitstream. (Actually the official Theora 1.0 encoder wasn’t all that bad, it was the implementation in ffmpeg that was terrible).
Thusnelda is a project to improve the encoder.
http://web.mit.edu/xiphmont/Public/theora/demo7.html
That project has, quite recently, started to get results. Recently this project has produced an encoder that gets results quite close to h264 at some bitrates when played back with the standard Theora decoder.
Thusnelda is now over half way (on a db scale) between where Theora 1 codec was an h264 performance. Much of the required catching up has already been done.
There is no more of “leaving Theora in the dust with ease”. Its “game on” now.
Edited 2009-05-11 04:00 UTC