Attempt at capturing the likeness of Bella. Was pretty difficult to get some good results, might have overtrained it a bit to get the face right, so expect lower flexibility. Might redo this in the future for a better version. Trained with 10 repeats and 20 epochs, used the sd 1.5 base model. Recommended values can depend on checkpoints, but probably between 0.5 and 0.8.