how does main prompt, and region prompts exactly work

[memedoggo]

when i run "active region layer only", is it
main prompts + region prompt,
or (main prompts) * 0.5 strength + region prompt

also if i run generation without checking region only,
do the region prompts still have some influence?

[Acly]

It's always "region prompt + main prompt", in that order.

Active region layer generates only the part of the image for the active region.
Otherwise generates the full image with all region prompts (affecting only their respective regions).

[memedoggo]

So, it only affects the area of the image it generates, while the prompts remain unchanged.

I thought it'd be great to have an option to reduce the batch strength for the main prompts because sometimes I want certain parts to have lower strength.

For example, I keep boots and footwear prompts because I want to maintain a universal prompt for every scene. If I want to create a scene without the lower body, reducing the strength of some prompts achieves this without completely removing them from each scene.

I also have a feeling that CLIP decoding works somewhat differently in the WebUI compared to Krita. In the WebUI, it generates such images even without reducing strength or removing partial prompts, whereas in Krita, it really tries hard to render all the tiny prompts.