I'm a long-time Theora fan, and though I certainly wouldn't complain over the inclusion of Theora, there are definitely reasons Theora wouldn't serve well for realtime video. The biggest thing to me is fault tolerance; lots of work has gone into VP8 to handle random packet loss -- force a little packet loss while streaming Theora and watch it turn to garbage until the next keyframe.
To industry (assuming no biases against libre software), the biggest issue will be hardware encoding support; for Theora, there's none -- not even multicore cpu support. For VP8, there's a little and it's getting better. H.264 wins for hardware support, and industry already favors it, but it fails any test of openness you can throw at it.
Theora's decoding performance is still considerably faster than VP8 device I've ever tested (I actually re-encode some videos to Theora for viewing on ancient machines), but it's losing its encoding speed edge every time an encoder update is pumped out for libvpx.
At this point, if I'm going to put in effort to support a libre codec, VP8/WebM's the one. If Theora sneaks its way into WebRTC, however, I won't complain.
![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
timeasmymeasure
Re: What about Theora
Date: 2012-08-10 11:13 pm (UTC)To industry (assuming no biases against libre software), the biggest issue will be hardware encoding support; for Theora, there's none -- not even multicore cpu support. For VP8, there's a little and it's getting better. H.264 wins for hardware support, and industry already favors it, but it fails any test of openness you can throw at it.
Theora's decoding performance is still considerably faster than VP8 device I've ever tested (I actually re-encode some videos to Theora for viewing on ancient machines), but it's losing its encoding speed edge every time an encoder update is pumped out for libvpx.
At this point, if I'm going to put in effort to support a libre codec, VP8/WebM's the one. If Theora sneaks its way into WebRTC, however, I won't complain.