Slashdot Mirror
Theora 1.1 (Thusnelda) Is Released
SD-Arcadia writes to tell us that Theora 1.1 has officially been released. It features improved encoding, providing better video quality for a given file size, a faster decoder, bitrate controls to help with streaming, and two-pass encoding. "The new rate control module hits its target much more accurately and obeys strict buffer constraints, including dropping frames if necessary. The latter is needed to enable live streaming without disconnecting users or pausing to buffer during sudden motion. Obeying these constraints can yield substantially worse quality than the 1.0 encoder, whose rate control did not obey any such constraints, and often landed only in the vague neighborhood of the desired rate target. The new --soft-target option can relax a few of these constraints, but the new two-pass rate control mode gives quality approaching full 'constant quality' mode with a predictable output size. This should be the preferred encoding method when not doing live streaming. Two-pass may also be used with finite buffer constraints, for non-live streaming." A detailed writeup on the new release has been posted at Mozilla.
Maybe now Google will use Theora instead of the patent-encumbered H.264 in their new HTML5 Youtube.
That is if the issues have been addressed.
Dirac strikes me as another codec worth following. It's available to all developers, high-quality, and in production use by the BBC during the Olympics (they said so in their Dirac promotional video). VLC has support for playing back Dirac streams. I'd guessing other players do as well.
I expect Theora and Dirac to be of interest to all who want high-quality free video codecs.
Digital Citizen
Hahaha, no. Just no.
Also hilariously wrong. Hell, one of the advantages (what few there are) of Theora its proponents like to bring up is that it takes less resources to decode than H.264. I have no fucking clue where you got this idea from.
Wrong again. There have been several comparisons between H.264 and Theora by the Xiph folks and they've all come out in favor of H.264. They've only tried to argue that Theora isn't really that bad. The problem is it is, and the only reason Theora didn't get utterly murdered in their comparisons is they've compared default Theora to default x264 and YouTube's H.264.
Default Theora is pretty much as good as it gets unless you want to set custom quantization/Huffman tables. Default x264 falls far short of x264 with its settings set for maximum quality, mainly because when you set them like that it's slow as fuck and most people will take worse quality over sub-1 FPS encoding. I don't know what YouTube uses or how they set it, but I seriously doubt a site that huge goes for the maximum possible quality.
Furthermore, Theora is simply inferior technology-wise to H.264. Theora-the-specification is far behind H.264 and it makes it pretty much impossible for Theora-the-software to ever be better than a decent H.264 encoder, as any improvements could simply be copied by the H.264 encoder (though it's more likely it'd be the other way around).
It is noticeably better than Theora 1.0, but remains noticeably worse than H.264 and will continue to be so.
I would rather that community based projects with low budgets distribute video using an absolutely free codec if the alternative is that they don't distribute at all because they can't afford the fees. If the quality is a little bit worse, but it's still fit for the purpose, and it's free, then it has more value than superior technology that is not affordable.
People shouldn't be using YouTube as their distribution mechanism in the first place. They should be using their own devices.
-1 Uncomfortable Truth
I made a few samples using the latest versions of x264, VC-1, and Theora, testing both offline VBR and real-time CBR encoding.
http://cid-bee3c9ac9541c85b.skydrive.live.com/browse.aspx/.Public/Theora%5E_1.1
Theora is defintely improved, but I see a lot of basis pattern throughout these samples. Theora would be well-served by a postprocessing filter. Theora's 1-pass CBR encoding definitely needs a LOT of tuning before it'd be viable for real-world content; I don't think we'll see it used effectively for live encoding this version.
My video compression blog
MPEG-1 is completely free, in most areas of the world, due to patent expiration.
Possibly - so long as you don't want any audio with your video.
It'll also put Theora to shame in just about every respect.
Unlikely - even the original VP3 can beat MPEG-1, despite its major flaws.
Encoding and decoding complexity is so low your digital watch could handle it, and h.264 offers practically no quality improvement at high bitrates, and only a small improvement at VERY LOW bitrates (what it was designed for).
Encoding and decoding complexity for MPEG-1 is... actually going to be quite close to MPEG-2. h.264 also offers quality improvements at *every* bitrate - due to CABAC (which provides better compression of the encoded data), better motion compensation that allows the available bitrate to be used more efficiently, and possibly even in-loop deblocking.
I wonder if it is time for me to reconvert all my HD movies again into this now. It takes hours to rip them, and even more to convert them. It could be worth it if it can be higher quality at the same bitrate or same quality at a lower bitrate. Some of my movies get a little messy in the high motion areas, but I didn't have much choice if I wanted all of them to fit on the same 1.5tb drive.