Topaz Video Starlight model question
When using Starlight precise or mini locally, do these models load only once and slow the first time, and processes subsequent batch videos faster or does it have to load model each job if same model is selected?
Lets say if I try to process 3 videos 1 min each 30 FPS, then I see job completion time is equal implying models are loaded every time or am I misinterpreting it?
When I use other open source diffusion models, its slow the first in ComfyUI but subsequent jobs are very fast.