Search options

It just doesn't seem like it is using GPU acceleration since my CPU usage is spiking so high when it's processing.  Is there an additional config I need to get vapoursynth to process on the GPU?

So I have it plugged into my pipeline now, doing frame-doubling via mpv.  I'm seeing very high CPU usage while interpolating.  Is this normal?

I have `--ovc=h264_nvenc` (not doing hevc as that's even more computationally expensive)
And `--hwdec=cuda-copy`

With those two settings, I should be using GPU for decoding and encoding, yet my CPU spikes to around 80% while processing.

Should I be using a different hwdec?  Is there any way I can reduce load on my CPU?  GPU usage is like 10% so it's doing something while transcoding, but I'd like to foist more of the processing load onto the GPU if possible.

Edit: Also, is there a way to force it to process at faster than realtime?  I have a 52 second file I'm interpolation, and the interpolation takes exactly 52 seconds regardless of whether I'm interpolating to 120fps or 48fps which makes me thing it's processing the video in realtime regardless of how much processor overhead it has.

Edit2: I should have posted the command I'm running for frame interpolation

mpv {input_file} --no-audio --no-sub-auto --input-ipc-server=mpvencodepipe --hwdec=cuda-copy --vf=vapoursynth:[E:\\\\ffff.py]:4:4 --ofps={new_framerate} --of=matroska --ovc=h264_nvenc --ovcopts b={new_bitrate},gpu=0,rc=vbr_hq,cq=25,preset=medium,profile=main,threads=4 --o={tmp_output_file}