On Thu, Jan 20, 2011 at 05:26, Shane M. Walton <[email protected]>wrote:
> Hello, > > I have created a libavdevice that can read in raw video frames from shared > memory, which allows me to run ffmpeg and benefit from all the features > built in. (I love this...) This works great for a known resolution that > is consistent. So my question is, can the application ffmpeg.c and the > format context and stream contexts used throughout react to a resolution > change determined in the libavdevice code? > > Ideally, the shared memory contains a description of the incoming frame, > and > when the description changes ffmpeg needs to be notified up stream to > recreate the video codec for the new resolution. This is that part I am > having trouble with. The AVInputFormat doesn't really provide a way to > modify the ost->st->codec and I doubt that I would want to modify that > directly... Is there a way to recreate the input stream context when > a AVInputFormat->read_packet() is called that would have the upstream parts > of ffmpeg recreate the codec or am I better off writing my own application > entirely? > > Here is the path of how I need things to work, the HD and SD video is > exclusive or not at the same time: > > Raw HD video -\ > SHM -> ffmpeg (via libavdevice) -> encode (h.264) -> > format (mpeg-ts) -> out... > Raw SD video -/ > > Thanks! > > Shane > that's something i'm very interested in as well, but in the case of MPEG-TS input. -- Aviad Rozenhek _______________________________________________ libav-user mailing list [email protected] https://lists.mplayerhq.hu/mailman/listinfo/libav-user
