<h3 id="brief">Brief</h3>Lora model of Belle from Zenless Zone Zero. Trained with 1536x1536 (IL,XL,Pony) and 1024x1024 (SD1.5) resolution, and enhanced with 3D model for best accuracy. (actually majority of the mataset is 3D model screenshots )The training dataset is categorized to further improve the flexibility, making my model easy to remove/change the costume.<h3 id="usage">Usage</h3>Trigger prompt is: belle \(zenless zone zero\)Additional prompts for Default Costume: letter hairclip, earrings, necklace, black shirt, sleeveless hooded jacket, single fingerless glove, single fingerless glove, bracelet, grey layered skirt, single black thighhigh, single orange sock, asymmetrical legwear, sneakersAdditional prompts for Delicate Daylight: letter hairclip, tassel earrings, puffy long sleeves, yellow sleeveless dress, white pantyhose, thigh strap, block heelsAdditional prompts for Summer Skies: letter hairclip, earrings, necklace, low twintails, black bucket hat, off-shoulder crop top, bikini under clothes, orange sports bikini, orange short, anklet, crocsPlease note that including all prompts will have the complete set of costume, remove one of more will remove the associated part of costume in most cases, so you know what to do.The model was well trained that the default costume can be removed/changed easily even with high lora weight, recommended weight is 0.8~1

Belle リン 3in1 (IL,Pony,XL,1.5)

Char-ZZZ-Belle-V1.safetensors

SD1.5 v1.0

v2.0-IL

Files with SHA256: b79ec440cec09fb0df01cc91a9c7923a978e26f48aab43311301396a96bd4dc3