New Ben model for ponydiffusion. works best with autism mix

trigger word: ben tennyson

neg: hairclip, blue clothes, Gwen Tennyson

might need adetailer for better result.

V2

better show-accurate model with slightly better omnitrix dataset.

now with less same ai-style base images!

use less quality prompts (ex:score [x] up) and try "parody"

to give your images show-accurate result.

+webui might have a hard time understanding omnitrix.
specific prompts such as wristwatch on left/right arm or omnitrix need to be added in.

from my personal experience, generating omnitrix in t2i and using inpaint to refine worked really well.

描述:

ver.1 has an issue with Ben having occasional Gwen-like features such as blue cat shirts or hairpins.

use negative prompts accordingly

I might adjust these issues later on. but no promises.

训练词语: