I have a laptop with both an AMD iGPU and an AMD dedicated GPU and I'm using shotcut which apparently uses the MLT Framework. It seems like shotcut is using the iGPU for HW acceleration based on encoding times. I can specify a target device in ffmpeg using the "hwaccel_device" parameter and it appears that shotcut has an option for passing parameters to MLT.
Is there a parameter that I can pass to MLT in order to specify the target gpu for ffmpeg to use? I guess I'm looking for the MLT equivalent to ffmpeg's "-hwaccel_device /dev/dri/renderD129"
The vaapi_device option may be helpful for others facing this issue. example usage: