in low vram situations and heavy offloading, 2nd pass might take longer than 1st pass, only for latent upscaling (distilled) - no idea how to "fix" or quantize latent space, but this would be very helpful. beside: great model! thank you very much!
· Sign up or log in to comment