I have been training the lora with new parameters to get a better result in v2. v2 works stably with WAI-NSFW-illustrious-SDXL, you no longer need to specify the eye color, the lora does not confuse it, I made some images for an example with the WAI-NSFW-illustrious-SDXL model and WAI-SHUFFLE-NOOB

Toxico Dannar (Futoku no Guild)

Toxico Dannarv3.safetensors

v1.0

v2.0

The sample images are made with the WAI-SHUFFLE-NOOB Eps model, with WAI-NSFW-illustrious-SDXL it also looks good, but I like how the eyes look with the noob model

Files with SHA256: c56647c271ed7828553993bb62a1ba6b3c4baa3cfed60bdd5f3923914574380b