1：When tegra_multimedia_API is used to loop out a video from camera_CUDA_V4L2, the probability of video will be fault, that is, the upper half and the lower half of the picture are out of sync, especially when collecting moving images.
2：Tegra_multimedia_api and video_CUDa_enc_main were used to encode a video along the way, and the encoded video also had fault phenomenon.
3：Using video_CUDa_enc_main example, if encoder’s output_plane, output_memory_TYPE is set to V4L2_MEMORY_DMABUF, fault will occur probabilistic; if ENcoder USES V4L2_MEMORY_MMAP, the encoded video will have horizontal stripe.
Could you help for this,thanks
Please refer to
It is to demonstrate 12_camera_v4l2_cuda + NvVideoEncoder. Looks close to your usecase.
will check it first