metadata
license: creativeml-openrail-m
license: creativeml-openrail-m
This model is dreamboothed on four concepts from yamanosusume
Prompts:
- aohina yuri kurauehinata yukimuraaoi 2girls
- yukimuraaoi girl
- kurauehinata girl
- aobakokona girl
Training details:
- Trained with TheLastBen's fast-DreamBooth notebook
- data set: 198 concept images, I cannot remember whether I used reg images or not (it seems not)
- learning rate 1e-6 for 15500 steps
- text encoder rate 10%
Problems:
- As always, it is hard to get interactions between multiple characters even though I explicitly trained for it.
- For some mysterious reasono, when generating images for aohina it can hardly get the eye colors right (maybe should train at clip_skip=2?).
Example generations (with inpainting):
