|
--- |
|
language: |
|
- en |
|
license: creativeml-openrail-m |
|
tags: |
|
- stable-diffusion |
|
- text-to-image |
|
- safetensors |
|
- disco-elysium |
|
inference: false |
|
--- |
|
|
|
# Harrier Du Bois |
|
|
|
Meet the [Harrier Du Bois](https://discoelysium.fandom.com/wiki/Harrier_Du_Bois). |
|
|
|
A disco elysium style Stable Diffusion model I trained a long time ago. |
|
Trained with [Disco Elysium Archetypes](https://www.artstation.com/artwork/6aAL8x) from [Aleksander Rostov](https://www.artstation.com/rostovjanka) |
|
and some painitng from [Craig Mullins](http://www.goodbrush.com). |
|
You might have seen it somewhere else. |
|
|
|
Based on a novel model we all know. |
|
|
|
Text encoder is trained and broken. **Very unwieldy** but interesting enough. |
|
|
|
## Usage |
|
|
|
No VAE. |
|
|
|
```txt |
|
a painting of a man in green suit by sks |
|
``` |
|
|
|
[swapping attention layers](https://gist.github.com/crosstyan/95d14111e8e1eeb3348ef947818b338f) |
|
might give interesting result. |
|
|
|
|
