Weight 1 work fine.
Note: By default, this LoRA will generate art style close to Booru/Illustration, use "anime screencap" if you want her to resemble the anime appearance more. Do not expect her rigging to generate properly, I removed most of it from the dataset because it kept adding random unusable garbage to pictures generated during previous testing and I had to retrain her twice for this matter.
(If you really wanna give it a shot try with "cannons", "rigging" or "machinery", "abyssal ship" but be my guest, do not expect consistency or perfect results).
Trigger: azlt3ster
Appearance: grey hair, long hair, hair ornament, yellow eyes, colored skin, pale skin
Helpers: anime screencap
Default Outfit: clothing cutout, cleavage, micro shorts, short shorts, black shirt, midriff, armored boots, fingerless gloves, criss-cross halter
Experimental (not guaranteed to give good results): weapon, floating weapon, rigging, cannons, abyssal ship