
This Lora was training on appx 120 images of varying sizes, dimensions, and resolutions at a variety of repetitions. I believe it creates the most accurate resemblance of her regardless of the model used. I used several models to create the examples.
Recommended strength is .85.



