Usa as negative: -0.4 to -2

Somehow a lot of my ANIMA images, especially when no humans are involved tend to have some kind of halftone or dithered texture on foilage or gradiants.

My Goal was to reduce this issue by prompting, but did not succeed - so I trained a LoRA on bad halftone textures... and not success either BUT it went out to be a pretty good detail enhancer when used as a negative LoRA.
I never made a workflow for Lora comparison, and I don't have the time for that right now. Look at the images - I provided the prompts and weights - the ones with plain or boring background are without the LoRA the others with.