about Tegra Multimedia API for cuda video conversion


I have question about video conversion by Tegra Multimedia API.

I see some samples, such as 03_video_cuda_enc. It needs a video file in YUV format to convert to the target h.264/h.265 format. However, I have many video files that are encoded in H.264 already. I just want to convert them into H.265. Is that possible I directly convert them into H.265 without a step in the middle in YUV format? This extra step does not make sense for me and waste time.

Also for 01_video_encode I am not sure it uses cuda acceleration or not. This example use CPU or GPU for encoding? For 01 and 03 video encoding, Which one is faster? I only run in silent mode that without show the video frames.

For transcoding, you have to decode h264 to YUV and then encode to h265.
Beside tegra_multimedia_api, you may consider to use gstreamer. Below post has a sample pipeline: