I’m in a project that needs video decoding in mjpeg format. I was really wondering about the best way to do this… I know that cuvid lib supports mpeg-1/2 and H264 decoding… but only that. Would it be wise to acquire each and every frame and treat it like jpeg decoding (decode jpeg files in GPU using cuda and map it as openGL texture, for instance)? or can I use the existent NVIDIA support for video decoding, somehow?
Is there any way to use or combine VDPAU + CUDA to do the same decoding? unfortunately, I don’t see many topics out there about vdpau + jpeg