Did anyone solve the multiple LoRA issue?!?
#89
by gottaloveit - opened
Did anyone solve the multiple LoRA issue?!?
When you try to use more than one LoRA the outputs suffer severely.
Does anyone think this will be solved once or IF the Base model is ever released?
Is it merely an issue due to the model being a Turbo / distilled?
Depends on the loras themselves
Yep _ and which is the lora made from ? i suppose FluX .. but then ? may be other model ?
from what is it done in facts and files ?
Don't know if it can help, but making some tests about this issue, I found that using a trained lora (not slider) with a negative strength of 1 (-1.0) did solve some issues.
I used a failed lora I made with Ostris's toolkit and the quality is improved.