V3:
High quality and enough versatile. Works wells with both photoreal and anime model.
Works best with not too small subject.
Suggested settings:
Width:>640
Height: >960
Hires-Fix: >x2 ✔️
Trigger word: hhayleyy
outdated V2:
I completely reworked the model. This time is trained in 2 different set of images for a total of 60. One ranked higher with 3x times the weight that the lower rank images.
Training done at 768x768 on the base model, this mean that you can use the Lora with evey model.
Trigger word: hhayleyy (not really required)
Suggested settings:
Width:>512
Height: >768
Hires-Fix ✔️
Description
This is my final version.
I can't get anything better with my current dataset. Seems incredibile but there are almost no high quality images of Hayley.
This version requires the use of Trigger Word