This LORA is a continuation, and hopefully an inprovement. The sequence has been HANNA-01 where the problem was that all training images were in the same hallway which continuously appeared in most outputs. HANNAH-02 used a white backdrop and the images defaulted to have white backgrounds.

HANAH-03 was created using the same woman, but the backgrounds vary hoping to solve the common or blank background of 01 and 02.

Testing will start now.. Fingers crossed.