>>89109245Baseline
Too high denoise in the upscale and try using somewhere around 0.3-0.4.
Do you have preview on? Just helps you see what you are generating before committing to a upscale.
Other than I agree with another try lowering cfg or use Noob 0.5. v1 knows a lot more but some training issues made it be pretty unstable.
>>89109834Euler a, I think is non convergent so it technically doesn't have one (Assuming I understood you right). 28-30 is what I generally see used for it.
