Amateur Adult Internet Star from 2007
Lora trained on 412 images of Liz Vicious. I am not her, find her civitai profile here
The prominent amount of the dataset is comprised of either red hair, or dual Red/Black toned hair. However, since many of the higher resolution images were later and had blonde and green hair, I usually will add at least green hair to the negatives on pony to get a more consistent hair result.
Natural language and booru tags work well, use the tag lizvicious.
Sample Prompt:
a photo of a lizvicious in a beautiful graveyard with a light fog all around, she has long dark red hair in an emo style with blue eyes and freckles, there are wiccan symbols and and candles and other gothic witch aesthetics, she has makeup on and is wearing a black choker with emo style, she is smiling with delicate soft pale skin
I recommend Piano Mix and Prefect Pony, however I'm sure others will work fine, but those are the main ones I've liked using with this resource. It was trained on the base pony model.
lizvicious, 1girl, emo hair, two-tone hair, makeup, long hair, jewelry, blue eyes, freckles, choker
If you have wildcard prompts
<lora:lizvicious_(PonyXL):1> lizvicious, 1girl, {emo hair, two-tone hair, | red hair,} long hair, blue eyes, freckles, makeup, small ???????, jewelry, {necklace, | choker, | spiked collar,} bracelet, rings, black nails, nail polish, smile, upper body,
Most common resolutions used during training
(1088 x 832), (832 x 1216), (832 x 1088), (1024 x 1024)
Note: I never managed to get the tattoo to work, if you find a way to get it, let me know in the comments.
Flux version, was trained on the same 412 images as the pony model, using booru style tags, however like other Flux models, still seems to respond well with natural language prompts.
1. 轉载模型僅供學習與交流分享，其版權及最終解释權归原作者。
2. 模型原作者如需認領模型，請通過官方渠道联系海藝AI工作人員進行認證。我們致力於保護每一位創作者的權益。 點擊去認領
