Stable Diffusion web UI

Introduction

I wanted to use an image generation AI in a local PC environment.
At present, Stable Diffusion seems to be almost the only choice.
There are several options for installing Stable Diffusion on a Windows PC.

• AUTOMATIC1111 Web UI: Famous
• NMKD Stable Diffusion: GUI
• InvokeAI: GUI
• ComfyUI: Node-based WebUI

This time, I will install AUTOMATIC1111 Web UI (Stable Diffusion web UI), which has a lot of information available, on my local PC.

Stable Diffusion web UI

Official GitHub (Stable Diffusion web UI)

A web interface for Stable Diffusion, implemented using Gradio library.

• Required Dependencies
  Python 3.10.6 and Git
• Operating Environment
  NVIDIA GPU (Recommended)
  While not mandatory, image generation will take a very long time without a GPU, making it impractical.
  (My home PC has an NVIDIA GeForce RTX 4060 Laptop GPU: 8GB VRAM)

Installation on Windows PC

There are two methods.

• Method 1: Unzip sd.webui.zip and install
• Method 2: Clone the Git repository and run webui-user.bat

With Method 1, Python 3.10.6 and Git will also be installed within the unzipped folder, so you don't need to install them yourself.

Method 1

1. Download sd.webui.zip from v1.0.0-pre (Figure 1).

Figure 1. Download sd.webui.zip

2. Unzip sd.webui.zip to any location.
3. Run update.bat located in the unzipped folder.
4. Once the update is complete, run run.bat, also located in the unzipped folder.
5. If your browser launches with the URL http://127.0.0.1:7860/ and displays the following screen, the startup is complete (Figure 2).

Figure 2. Stable Diffusion web UI started

6. Enter a prompt and press the Generate button to generate an image (Figure 3).

Figure 3. Image generated

Method 2

1. Install Python 3.10.6 (64-bit) and Git.
2. Clone https://github.com/AUTOMATIC1111/stable-diffusion-webui.
3. Run webui-user.bat located within the cloned repository.
4. Thereafter, it is the same as Method 1, and the browser will launch at http://127.0.0.1:7860/.

How to Quit

You can either press Ctrl + C in the command prompt or close the command prompt.

Additional Notes

• You can also enter Japanese in the prompt, but the accuracy is poor, so it's best to enter in English.
• The Stable Diffusion checkpoint in the upper left refers to the image generation AI model (checkpoint: pre-trained model file). The default v1-5-pruned-emaonly.safetensors is a lightweight model and not particularly high quality. The generated images vary greatly depending on the checkpoint. Checkpoints can be obtained from Civitai and other sources.

References

Install and Run on NVidia GPUs