Hi All. I'm trying to get a V4L2 M2M driver sorted for the Raspberry Pi to allow access to the video codecs. Much of it is working fine. One thing that isn't clear relates to video decode. Do the compressed formats (eg V4L2_PIX_FMT_H264) have to be framed into one frame per V4L2 buffer, or is providing unframed chunks of an elementary stream permitted. The docs only say "H264 video elementary stream with start codes.". Admittedly timestamps are nearly meaningless if you feed in unframed data, but could potentially be interpolated. What does other hardware support? I could handle it either way, but there are some performance tweaks I can do if I know the data must be framed. Thanks in advance. Dave
