View different layers, observe U-Net feature maps. Image generation by giving different prompts for each block of the unet: https://note.com/kohya_ss/n/n93b7c01b0547
f328a810b9
|
||
|---|---|---|
| images | ||
| javascript | ||
| scripts | ||
| .gitignore | ||
| LICENSE | ||
| README.md | ||
| style.css | ||
README.md
An extension for stable-diffusion-webui that adds a custom script which let you to observe U-net feature maps.
Example
Model Output Image:
Model: waifu-diffusion-v1-3-float16 (84692140)
Prompt: a cute girl, pink hair
Sampling Method: DPM++ 2M Karras
Size: 512x512
CFG Scale: 7
Seed: 1719471015
U-net features:
- IN00 (64x64, 320ch)
Let the feature value is v, larger |v| is white, and zero is black.
step 1
step 10
step 20
- OUT02 (16x16, 1280ch)
step 20
- OUT11 (64x64, 320ch)
step 1
step 10
step 20
Color map mode:
Red means the value is positive, and blue means the value is negative.
