This version was trained with the default setting on OneTrainer using the same images as v1.0. I think it offers a more accurate and consistent face from image to image, using a lower LoRA strength—you be the judge, and please let me know what you think!
I suggest using This_Rock's ?????? size slider/Small ??????? (FLUX) at a -1.0 setting to achieve a more realistic "Kate Hudson" body.



