A LoRA that tries to output images of a character looking at their skeleton reflection. Sometimes the skeleton have the features of the character, but it's not consistent.

Thanks to advanced mode, it kinda works with Illustrious checkpoints, but generation takes forever, with high failure rates

All training images were in the same pose, so it's limited in this regard. The training data is available to anyone who also want to take a shot at this (maybe even improve it idk)