stablediff

Toward Generations of New Gacha Images for Better h(hyeok)-index

from https://github.com/nonmetal/stable-diffusion-webui (deprecated)

Style Transfer with Modified Layer Structure

1. Generation from base model

looking at viewer, portrait, 1girl, blonde hair (default negative prompts)

1. Hypernetwork training with Modified Layer Structure

`looking at viewer, portrait, 1girl, blonde hair` *(default negative prompts)*
`nn.Linear(), depth == 2` (base)	`nn.Linear(), depth == 6`	`nn.conv1d(), depth == 2`