> (and Google won't host video in Theora, as the video is too big at acceptable quality levels)
Chris DiBona of Google made that claim about Theora on the WHATWG mailing-list, but it turns out that Theora can produce nicer video than H.264 at the bitrates YouTube delivers:
Of course, it's quite possible that Theora is more processor intensive than H.264 for an equivalent bitrate, but in raw quality-per-byte Theora is perfectly usable for YouTube-quality video.
Sorry, no. Theora is better than H.263, which YouTube currently uses for compatibility with older Flash players. This does mean that Chris DiBona was wrong, but Theora is still worse than H.264 (which YouTube uses for all but the lowest quality clips).
Check out the last graph on the page. As they point out, PSNR isn't a 100% unbiased way to compare codecs. However, the graph, made by the theora developers themselves running the bleeding edge encoder on content of their choosing, clearly shows that Theora requires 50% larger files to produce the same PSNR as H.264.
I should never have put in that reason into the parenthetical. I did not want to get into the "no it's not," "yes it is" argument over Theora vs H.264.
The point is, neither Theora nor H.264 can actually be standardized on for now, for a variety of reasons. Maybe the reasons are good, maybe they're bad, but several of the most significant companies involved refuse to support one or the other, and can't be convinced otherwise.
A new, royalty free, open standard codec might help break the deadlock. As I said, there are still hurdles, like the hardware support issue, and the substantial cost to Google when it's already invested in H.264. But I'm just not really seeing Google in the business of selling proprietary codecs, so I can't really see why else they would buy On2.
Did you even read that comparison before posting about it?
It explicitly disclaims that the only reason Theora won was because of Youtube's utterly awful H.264 encoder--which they have since replaced with x264 for exactly that reason.
gmaxwell, one of the Theora devs and the author of that post, has personally requested that I bonk every single person who makes any claim as to Theora being nearly as good, equal to, or better than H.264, because such claims create expectations which they know they cannot possibly reach. The single worst thing people can do for Theora is to market it based on false claims.
Chris DiBona of Google made that claim about Theora on the WHATWG mailing-list, but it turns out that Theora can produce nicer video than H.264 at the bitrates YouTube delivers:
http://people.xiph.org/~greg/video/ytcompare/comparison.html
Of course, it's quite possible that Theora is more processor intensive than H.264 for an equivalent bitrate, but in raw quality-per-byte Theora is perfectly usable for YouTube-quality video.