LoRA trained from 50+ finely selected images on the internet. Best to use in combination with Anything V4.0.
Suggested LoRA weights: 0.5 ~ 0.9.
Suggested Img Size: 512x512/512x800/768x512
Tips:
Use in combination with control net plugin on in-game screenshots yield great results:
Initial draft may look underwhelming, please don't forget to use hires.fix to upscale your image (Denoising strength: 0.5 ~ 0.7)
(Minus the fingers :D)
Prompt words:
rosaria_\(genshin_impact\), bangs, crown, elbow_gloves, bare_shoulders, fishnet_legwear, cape, weapon,
Extra Fun Prompts:
Kimono, realistic, spear, weapon
Description
LoRA model trained from Anything V4, which should be compatible with Stable Diffusion v1.5.