Tsukushi from DOAXVV
trained on data with head view only : back, front, diagonal three-quarter.
trained without : eyeglasses, hair-clip, body.
no "eyeglasses" in this training data, this lora will not interfere if you mixed with other eyeglasses type lora.
no "body" or "clothes" in this training data, image will not biases into a certain outfit or bodytype, training data only on head and hair.
maximum flexibility, can generate all type of eyeglasses, and body type. without lora bleed into outfit or hair accessories.
future improvement V1.1: "short brunette hair" maybe replaced, the "short" may leak into bias generating short skirt, or dress.