In the TVAI app on windows, i can set the AI processor to CPU. Also I can manage the models and mark and download the ones that I want for CPU usage. As a result the app renders (slow) on the CPU only.
Can this be replicated by command line usage in python? this ffmpeg always ends up GPU, even I download the CPU models in model managment:
"cmd": [
"S:/_apps/topaz/Topaz Video AI 5/ffmpeg.exe",
"-y",
"-loglevel","error",
"-stats",
"-nostdin",
"-hwaccel","none",
"-flush_packets","1",
"-ss","00:01:54.000",
"-hide_banner",
"-i","C:\\DUST_VIDGRAB202312\\2023-12-15_21-16-13_R06.00_cropped.mkv",
"-to","00:01:55.499",
"-sws_flags","spline+accurate_rnd+full_chroma_int",
"-filter_complex",
"tvai_fi=model=chf-3:slowmo=1:rdt=0.01:fps=30:device=-2:vram=0.8:instances=1",
"-c:v","prores_ks",
"-profile:v","2",
"-vendor","apl0",
"-quant_mat","hq",
"-bits_per_mb","1350",
"-pix_fmt","yuv422p10le",
"-movflags","frag_keyframe+empty_moov+delay_moov+use_metadata_tags+write_colr",
"-an",
"-map_metadata","0",
"-map_metadata:s:v","0:s:v",
"-metadata","videoai=CLI: Framerate changed to 30 using chf-3 replacing duplicate frames. range: part_d2.0_s0114.0_e0115.5",
"C:\\DUST_VIDGRAB202312\\parts_d2.0\\2023-12-15_21-16-13_R06.00_Rai30.00_part_d2.0_s0114.0_e0115.5_cropped.mov"
],
Why do I want this? Since 2 months my nvidia GTX 980M keeps crashing. My CPU supports POPCNT. I have tried reinstalls etc. I still want to sustain the machine.
Thank you, Michael