Style lora of wish72's work. Really likes indoor locations. I trained it mainly for the body proportions, but at 0.9 and 1.0 it gets the faces pretty close. I suggest using it on 0.75 with a tag w1sh72. You can also add 3D, realistic, sharp focus. Really likes to add more fingers. Still trying to figure out how to not overfit the loras and make them more flexible even at 1.0 strength... Works a bit better with 3d checkpoint.

Wish72 style

First version, could be less flexible than expected

Files with SHA256: c9aae7ccbd45fa9789b7984af837f504c023a3e7be5d8c6af6750dc43aec8db1