100% experimental lora.
The project dates back to early 2024 but only came to fruition in 2026.
Initially, I used test images from the Real Time Generator, which I carefully kept.
In the end, I finished using CivitAI because it was faster.
Trigger words: (girl in hell), (1girl), (naked), (surrounded by hell monsters), detailed, dark color, realistic screencap
Important prompts (the atmosphere is very muddy, with blood) : (mud:), (blood:), (bloody skin:), (dirty:), (dirty body:) , (dirty skin:), dirty monster, (skull monster:), (skeleton monster:), (worm monster:), (tentacle monster:), (tentacle worm monster:), (zombie monster:), empty eyes, tears, cry, drooling, (mid-thigh mud:), Hell cave...
Depending on the long-term results, a new version will follow. and a tentacle monsters version (there are some in the images), is in the works.
The style matches the images, as does the atmosphere, but sometimes, curiously, the monsters are absent, or nearly so.
I believe it is necessary to run various tests and trying out different models.
Some models are more susceptible to mud, and others to blood.
Strenght: 0.7 -> 1
