README.md

sdxl-demos

Python demos for testing out the Stable Diffusion's XL (SDXL 0.9) model.

Requirements:

• Hugging Face programmatic access
• Sign Stability AI's SDXL licensing agreement on Hugging Face
• Python (3.10)
• Nvidia GPUs or Apple M1 / M2 chips.

Installation

Create Python 3.10 virtual environment

git clone https://github.com/newtonapple/sdxl-demos
cd sdxl-demos
python3.10 -m venv ./venv/
source venv/bin/activate

PC Installation

pip install -r requirement.txt

Or with development dependencies:

pip install -r requirements/dev.txt

MacOS Installation

pip intall -r requirements/macos.txt

Or with development dependencies:

pip intall -r requirements/dev-macos.txt

Running Demos

text-to-image

CLI:

python txt2img.py 'an italian chef making pizzas' images/pizza-making.png

low VRAM mode:

python txt2img.py 'an italian chef making pizzas' images/pizza-making.png -l

Gradio WebUI:

python txt2img_webui.py

low VRAM mode:

python txt2img_webui.py -p 7860 -l

Open http://127.0.0.1:7860 in your browser.

WARNIN: width & height must be multiples of 8 or the WebUI will crash.