Honestly it LoRa trained on vegetation , buildings and 3DCG girls , and psychadelics. Right now model is a bit sparse but I'll steer it towards McGee's Alice in Wonderland + similar styled characters in next training iteration.

Intended to be used for drawing reference. The training images are provided in showcase. This an early trained version. Trained on 300 images , over 10,000 steps over 5 epochs , 6-7 repeats per image.

Set to Learning rate 0.0001 constant.