A LoRA designed primarily for producing scenes containing both curly brace and quote from cave story though it can do solo scenes as well.
use notes: best paired with one or more style loras, weight 1 was found to be sufficient, austismmix base model.
why did I make this: I wanted to see if a lora can be used to understand more than one character at a time without getting them confused. It mostly can.
Description
initial release
FAQ
Comments (3)
oh, that's very interesting
Question: When you were giving the LoRA images to train on, did you happen to use any images where they were tied back-to-back. Like when Quote was carrying Curly underwater or in Running Hell, cause I have a theory that I'm tempted to test and would appreciate some insight on others' experiences before I make an attempt for something that might screw myself over.
No, having gone through the public image boorus I'm not sure such a picture even exist. For training I used 51 images featuring either one or both characters the majority featured both.