
Seraphina, an OC a friend of mine made for a game he's working on. Trigger word is seraphina
We asked ChatGPT to generate descriptions for characters of an RPG game and broke down those descriptions into tags.
This lora is trained on images that were AI generated with prompts we normally use to make this character. Those prompts are "female with long wavy platinum blonde hair wearing a blue cloak and a chemise" and "Grey eyes, blue robe, platinum blonde hair, long hair, corset, belt, holding staff".
Tested on 0.4-0.85 weight. Higher weights can get decent generations of her "canon" look even with a prompt as simple as "female, solo, seraphina, upper body"
Example prompt 1: seraphina, solo, skirt, shirt, long sleeves, dress, closed mouth, thighs, belt, wide sleeves, corset, side slit, holding staff
Example prompt 2: seraphina, solo, cleavage, open clothes, hood, white dress, open jacket, hood down, blue jacket
This lora is mostly just an experiment, as it is the first time I train a lora with a dataset made 100% of AI generated images, there are no plans of making another lora of similar nature in the foreseeable future.
描述:
训练词语: seraphina
名称: Seraphina_OC-60.safetensors
大小 (KB): 36981
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success