>>2650
>yup, 0.33 iirc. Did you take a guess or do you have THAT good of an eye for that?
It was the guide's default, but I also a guess given the almost-perfect but still imperfect quality on certain parts when zoomed in.
0.33 is a good stable value. For max detail 0.4-0.5 is the spot.
The problem with 0.4-0.5 is it almost always creates random errors.
I get around this by plotting 3 or more gens (x/y/z plot -> denoise at 0.5,0.3,0.2) and then compositing: meaning I soft crop the good areas on the error ridden high denoise gens by erasing a layer on top of another.
You typically want this on high attention areas like face, and also small sections where it omits detail, can be nipples, fingernails, labia-clit, a small object, stuff like that.
This is more an OCD perfectionist thing. I don't think 99% of people will scrutinize your work pixel by pixel.

>Is there something in the pic that leads you to believe I may not have had it ticked? :)
No, just making sure since it's a new tool for you, Mutltidiffusers without mixture of diffusers is just slightly worse

>Still, even I can see that the overall quality is improved, despite me usually beeing rather blind to this sort of thing. (So again many thanks for showing me the light! :D )
You're very welcome

>On a related note, by doing this pic I came to notice, that apparently Adetailer is working far worse in Forge in some cases for me. At least concerning hands. I tried for hours but no matter what I tried, the output never came even close to the results from a1111.

I don't think its behavior should be any different on forge. Forge handles loras differently but the tools should be exactly the same.
Adetailer hands was always shit for me, even on 1.5

Personally I don't use adetailer anymore, face or hands. I just let multidiffusers enhance face and hand detail automatically. It does a good enough job and being a separate part of the process I don't need to worry about adetailer screwing up the gen, something that can always happen no matter the model