• Stable Diffusion Intro:
  • Text-to-image latent diffusion model
  • Created by CompVis, Stability AI, and LAION
  • Trained on 512x512 images from LAION-5B database subset
  • Uses frozen CLIP ViT-L/14 text encoder
    • part of the CLIP
  • 860M UNet and 123M text encoder
  • Lightweight, runs on consumer GPUs

- Try Stable Diffusion from Huggingface - install some packages ```bash pip install -r requirements.txt

# if you are in China
pip install -r requirements.txt -i http://pypi.douban.com/simple/ --trusted-host pypi.douban.com
```

• Pipeline
  • end2end
  • pretrained model on huggingface:
    • CompVis/stable-diffusion-v1-4。 512x512
    • runwayml/stable-diffusion-v1-5. 512x512
    • stabilityai/stable-diffusion-2-1-base. 512x512
    • stabilityai/stable-diffusion-2-1. 768x768
  • for a faster inference and lower memory usage, use fp16, also pass torch_dtype = torch.float16