>>106362913
>* 0.15
why tho
try 0.125 or 0.25
(8x * one of those is likelier to give you an even number with one of the resolutions divisible by 64)
>8x-super twice
oof
do 8x on first pass and something softer on second, or some images will look like cancer
and really you should have a second ksampler between at very low denoise and 8-12 steps
takes longer but so worth it