Works better
For preview 3. Works fine at 1.0 strength. Euler_a > other samplers; normal > other schedulers. Unfortunately, the model does not draw film grain well, and I can't solve this problem.
