NVIDIA/GMAT

How to scale and convert frame which pix_fmt is AV_PIX_FMT_CUDA

Opened this issue · 0 comments

bug-developer021 commented

https://docs.nvidia.com/video-technologies/video-codec-sdk/12.0/ffmpeg-with-nvidia-gpu/index.html#hwaccel-transcode-with-scaling

According to https://github.com/NVIDIA/GMAT/blob/main/ffmpeg-gpu/libswscale/cuda/swscale_cuda.c#L34, supported_fmts do not include AV_PIX_FMT_CUDA， which my ffmpeg hw_decoder px_fmt is AV_PIX_FMT_CUDA, how to scale and transform px_fmt to RGB, then to host memory?

As far as I know AV_PIX_FMT_CUDA is the NV12 format on the GPU, which is YUV_NV12， So, how to support AV_PIX_FMT_CUDA format when use libgpuscale