metadata
license: creativeml-openrail-m
Prompt with "francat"
Training details:
- Trained with TheLastBen's fast-DreamBooth notebook
- data set: 225 concept images + 224 custmized reg images
- learning rate 2e-6 for 5500 steps
- text encoder rate 15%
Problems:
- The generated images can get blurred easily is often quite dark, especially when anime screenshot is specified.
- Cannot get the clothes and the sword properly. This may require tagging the clothes separately.
- The model can never put the cat tail at the right position.
Example generations (with inpainting):
