On 17 April 2016 at 03:40, Michael Niedermayer <mich...@niedermayer.cc> wrote:
> > is this a spec compliance or a profile compliance issue or something > else ? > > from just the Commit message above it sounds a bit like a profile / > hw decoder limitation, in which case this patch with a different > av_log() message should be ok ... > > It's a hardware limitation. Basically, SMPTE ST 2042-2:2009 (a separate specification from the VC-2 specs) defines levels 1-7, an each level has a few associated base video formats (e.g. 6 for 4K and 3 for 720/1080). So each hardware decoder which claims to support a level must support all video formats associated with that level. The VC-2 specifications say nothing about restricting the level. Our decoder doesn't care about the level since Dirac was a general purpose codec and so it supports everything. BTW the encoder hasn't been tested on hardware, and I've only heard of legends about such existing. The spec isn't well written so I'm pretty sure turning on huge slice sizes would crash hardware decoders since they'd have no memory. So the level value doesn't do much at all to ensure hardware can still decode. _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
