hafiidz / latent-diffusion

High-Resolution Image Synthesis with Latent Diffusion Models

Fork for use with sd-webui

• If you're following the steps outline via Installation Link, there is an optional steps to load Latent Diffusion Super Resolution (LDSR)
• This repo helps minor updates to automatic download_mode.bat and keep things compatible with sd-webui
• download_mode.bat - will download the required model files and place them under experiments/pretrained_models

LDSR

1. Git clone this repo Hafiidz/latent-diffusion into your /stable-diffusion-webui/src/ folder
2. You can do this by navigating to /stable-diffusion-webui/src/ folder in your VSCode terminal, or via running git bash or other relevant command prompt

3. Run git clone https://github.com/Hafiidz/latent-diffusion.git
4. It will take a while to download and once the cloning is completed, you will see the following messages:

5. Next, run /stable-diffusion-webui/src/latent-diffusion/download_model.bat, by double clicking it, to automatically download 2 relevant files project.yaml and model.cpkt
6. The file will take a while to download, once done, navigate to stable-diffusion-webui/src/latent-diffusion/experiments/pretrained_models/ and confirm the two files are there

7. (Optional) If the two files are not there, you can manually download them:
   1. Download project.yaml and model last.cpkt.
   2. Rename <date>-project.yaml to project.yaml and last.ckpt to model.ckpt
   3. Place both under stable-diffusion-webui/src/latent-diffusion/experiments/pretrained_models/
8. Follow the discussion or raise a new issue here.

Link to Latent diffusion details

• Original Readme
• CompVis