This LORA was trained on images containing both Rogue and Kitty which were captioned with a mix of natural language and tags and should be able to produce scenes with both of them. The appearance of the characters is primarily based on their depictions in Project Evolution and Rogue-like: Evolution. The prompting style needed to generate them reasonably consistently is shown in the example images. Generally one run-on comma-free sentence explaining what each character looks like and what, if anything she is wearing should work most of the time. Multiple sentences increase the chance of unwanted contamination. Consistency can be increased by specifying the position of each character, e.g. "the girl on the lower left is k1tty with blue eyes and brown hair in a ponytail,... the girl on the upper right is r0gue with grey eyes and short brown and white multicolored hair,...".