Hi, I'm implementing dirac decoding (with schroedinger) for libquicktime and gmerlin-avdecoder.
I think I found 2 problems: 1. From the mapping spec, section 5.3 "Definition of a sample containing Dirac": The end-of-sequence should *never* be in an own sample. Because every demultiplexer will treat it as a video frame, which will fail (i.e. total duration is wrong and the last "frame" cannot be decoded). I would either allow the end-of-sequence appended to the sample containing the last frame, or be omitted completely (with schroedinger it's not really needed). 2. The test file http://diracvideo.org/download/test-streams/quicktime/sage-640x360.mov has no ctts atom, but decoding_order != display_order. Without a ctts atom, correct pts handling becomes as brain-dead as for AVI. The ctts atom should become mandatory for streams with B-frames. Please cc me, I'm not subscribed. Burkhard ------------------------------------------------------------------------------ _______________________________________________ Schrodinger-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/schrodinger-devel
