Under the model page with LoRA of one mighty deer japarn7 other curious deer asked the question, “Did you try to use only High? I finished my coffee and realised that I had only ever used it in High and had never tried using in both High and Low at once.
So there is the test of Squat Bounce (GoodThing) Dance LoRa. I tried making a video with only High noise LoRA, with High noise LoRA, and with the same LoRA in Low noise.
So there are two videos. Both videos were made with the same seed, 8 steps (4 high, 4 low) and CFG 1.5.
Video with the same LoRA in high and low noise.
Video with LoRA only in high noise
I can only see a difference in facial movements and, perhaps, in some minor details. When watching the videos separately, it seems to me that the videos with LoRA in High and Low noise have become more dynamic. But when I watched them together (placed side by side on the screen), I could no longer see the difference.