I'm creating a video from scratch using moviePY. I am generating all the required frames, adding required audio (including a voiceover and background music), adding a logo and finally writing the video file in 4K.
Everything works fine, except the write_videofile takes a lot of time.
I have read many related posts which mentions using the right codec etc. I have NVidia card, so tried both h264_nvenc and hevc_nvenc. Quality of the output video dropped with the first one, so I'm sticking to hevc_nvenc. I'm using below line to write the file.
video_clip.write_videofile(targetfile, codec="hevc_nvenc", threads=32, fps=24)
What I have noticed is, it does seem to be using the gpu, but very little. Compared to this, when I run stable diffusion or vegas rendering, it uses gpu a lot more.
That's why I think there is definitely some scope of improvement here. If you see below screenshot, when the video file write starts, the gpu utilisation increases a tiny bit, but it can take a lot more I think, isn't it?
I can try various parameters that I've seen in other threads, like logger=None, progress_bar = False, ffmpeg_params=['-b:v','10000k'] etc., but they are not going to improve gpu utilisation in any shape or form. I've been wondering what am I missing.
Any ideas or suggestions please?
