Generally I've noticed that the models I've been making on cetus mix blend very poorly with the new loRa s I'm making on 1.5 or anything v4.5-5, in fact I was editing the dataset, for some reason it had extra characters in it (I was very incompetent 5 months ago) that could cause the hair color to turn blue

Also the cetus mix based loRAs looks pretty dull, and this one looks the same as any other based on 1.5

-More shadows

-Better eyes (not as matte)

-The eye color itself looks better.
This loRA is much more sensitive to hair color, so if you have problems with it use version 1 or write red(orange)hair in the prompt