Ffmpeg and batch processing

I understand that batch processing in the UI does not work. So i am fine with using the terminal and create myself some batches.
(please correct me if i am wrong below, i would love to be wrong).
The problem i am having is with the bundled ffmpeg. I need to use that to apply the ai models for the upscaling etc. But that creates gigantic output files (upscaling a low quality SD video to 4K went from a 300mb file to a 17GB file).

I tried to modify the command line to use -CRF or -b:v - that they seem to be getting ignored. So my current process is to do the upscaling with the bundled ffmpeg, and then have a second pass to compress the video with the official ffmpeg. That seems like a bit of a waste of time.

Am I doing it wrong?


1 Like

Oh, and while i am having you reading this: is there a way to use more cpu power? the default command line creates me around 3-4 fps, and my machine is more or less sitting idle (using -threads etc does not seem to do a thing ). I have a mac studio superduper, there is plenty more cpu than the 15% that is getting used when I am running a conversion.


To answer myself, the bitrate seems to work, i just did not find the original command because it had a 143.234234234M value for b:v and i just skimmed past it. Running it right now and it seems to create a 4K output with a 4K bitrate for v. I will know in a couple of ours if that is actually working.

I can’t seem to get my CPU usage above 10 to 15% whatever I do, it basically uses only one core/2 threads of my 16 available.

GPU usage is higher than v 2.6.4 though, but I don’t know if that is because it’s actually doing more useful work or is just less efficient at what it does.