On Mon, Nov 30, 2015 at 8:21 AM, Nicolas Derouineau <nicolas.derouin...@vitec.com> wrote: > Hello > >>you should look at avframesidedata > > Actually the issue is that the green metadatadata are concerning the upcoming > frame (They are used to predict the upcoming complexity). > > The avframesidedata are concerning a frame which is already decoded (so it's > not useful in the case of greenmetadata). > > We made a proof of concept using a home made parsing function (replacing > av_read_frame() by a custom one), but we are thinking about a more portable > solution.
I am definitely not familiar with what the green metadata is or does, but from the sound of it I still think it is suited for AVFrameSideData: it's data which is not essential for decoding, it's frame bound, and applications outside lavc might need to act differently depending on its contents. Even if this green metadata is used within libavcodec and concerns the upcoming frame, there is still a way to make it work: I believe that frame side data in ffmpeg are ref counted (meaning you can just add a reference to the side data rather than allocating/copying data), put it in a queue of green metadata and apply it when the upcoming frame is decoded or it's about to be decoded. You would have to be careful only in frame reordering, but the rest should be pretty simple. Vittorio _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
