I was just looking at http://svn.xiph.org/experimental/derf/theora-exp/README and this is what it says about the encoder, <snip> The encoder is highly experimental. It currently generates valid bitstreams using ALL of the decoder features supported above. However, it is also currently broken, producing pretty horrible quality files with high bitrates. This should be tracked down and fixed before too much longer, but it hasn't happened yet. </snip> That doesn't encourage me to try it. Regards, John
A number of bug fixes have gone in recently, which improve things greatly. I haven't updated the README because I haven't done sufficient testing to ensure that everything is truly fixed. The psy model probably still needs some work. It can be disabled, but currently only by editing some code. In any case, it's still probably not in a position where running a bunch of tests and posting pretty graphs on a web page somewhere or something similar is worth the time. The decoder, however, has basically stabilized. If that does not produce bit-identical output compared to the mainline decoder, I want to know about it. It should also give decent speed improvements over the mainline decoder, especially at higher resolutions. I haven't spent (and probably won't spend) a lot of time optimizing it or adding SIMD assembly support.
John Kintree wrote:> gcc -I../include -I/usr/include/SDL -D_REENTRANT -g -c -o > objs/player_example.o ../examples/player_example.c > ../examples/player_example.c:53:17: SDL.h: No such file or directoryThe player example requires SDL. There's currently no autofoo, so if you have it installed somewhere other than the default path, you'll have to edit the hardcoded strings in the Makefile.