Attempt at a style based paintings of the underworld. The detail is poor but it does create interesting compositions. It's based on a few paintings of the underworld by classic painters. Source images were 512x512. Mature as the paintings were pretty gruesome. I don't think what the trained images depict make the lora do what it does. It's more to do with the composition of the images and how they are diffused.

Trigger word is underworldstyle. You have to specify the trigger word and the lora tag in the prompt. Different weights have different effects.

Example prompt:

warrior, underworldstyle, <lora:underworldstyle:1>