README.md

---
license: creativeml-openrail-m
---
# 📜🗡️ Silhouette/Cricut style
This is a fine-tuned Stable Diffusion model designed for cutting machines.
Use **silz style** in your prompts.

### Sample images:
![silz.jpg](https://huggingface.co/PiyarSquare/stable_diffusion_silz/resolve/main/silz_characters.png)
![silz.jpg](https://huggingface.co/PiyarSquare/stable_diffusion_silz/resolve/main/silz_famous_people.png)
![silz.jpg](https://huggingface.co/PiyarSquare/stable_diffusion_silz/resolve/main/silz_animals.png)
![silz.jpg](https://huggingface.co/PiyarSquare/stable_diffusion_silz/resolve/main/silz_places.png)
![silz.jpg](https://huggingface.co/PiyarSquare/stable_diffusion_silz/resolve/main/silz_prompted.png)
Based on StableDiffusion 1.5 model

### Training
Made with [automatic1111 webui](https://github.com/AUTOMATIC1111/stable-diffusion-webui) + [d8ahazard dreambooth extension](https://github.com/d8ahazard/sd_dreambooth_extension) + [nitrosocke guide](https://github.com/nitrosocke/dreambooth-training-guide).

82 training images at 1e-6 learning rate for 8200 steps.
Without prior preservation.

Inspired by [Fictiverse's PaperCut model](https://huggingface.co/Fictiverse/Stable_Diffusion_PaperCut_Model) and [txt2vector script](https://github.com/GeorgLegato/Txt2Vectorgraphics).