4 ms·

The Theora/VP3 code cannot be substantially improved without breaking backwards compatibility. While there are some gains to be made with better encoders, some

by e1ven 17y ago

The Theora/VP3 code cannot be substantially improved without breaking backwards compatibility. While there are some gains to be made with better encoders, some of the choices, such as small fixed block-size, permanently hobble the codec. There are those far more qualified than I to go into the technical details (I think there was an article by one of the FFMPEG devs), but the situation is not analogous to early versions of Webkit, Linux or Mozilla. This is not a situation where it can substantially improve over time, which is one of the major advantages of an Open Format.

ZeroGravitas 17y ago
What is stopping them from revving the spec and breaking compatability? Is it all those hardware implementations? Clearly not. And existing content would still play just newly encoded material wouldn't play on old encoders. Very similar to H.264 profiles, where Main and High profiles won't play on iPhones. I would suggest they haven't done this yet because they are making rapid progress without having the need do so. The longer they leave it the more they can improve when they break either by taking advantage of more expired patents or just more experience. If the argument is that Theora can't be revved (for some unspecified reason) and that limits progress then that argument should be made so it can be challenged. It does not follow directly from anything said in the article.
- wmf 17y ago
  What is stopping them from revving the spec and breaking compatability? Basically that's Dirac, except they started from scratch rather than improve incrementally from Theora. There was also Sun OMS Video (which has probably been canceled) that attempted to incrementally improve H.261 in a patent-free way.