Trained with cinematic film stills of the sequels.
So it is intended to reproduce the likeness of Daisy Ridley in her role as Rey.
Her name "rey" is the main trigger for this Lora.
Trained it with the following trigger as well for some of hher features:
scavenger
battlestick
jedi
blue lightsaber
hood
Description
FAQ
Comments (2)
I noticed that in the training data you have "desert" written as "dessert".
Also the captions used are very short, perhaps using LLaVA to augment the captions could help make the loras more versatile.
Sounds interesting.
Is there any good guide out there how to utilize LLaVA for captioning?