I know a lot of you are interested in AI-generative art, and sometimes I suggest Stable Diffusion. Today, I want to share something to help you start your Stable Diffusion journey!!!
Table of Contents
- Table of Contents
- What is Stable Diffusion?
- Why use Stable Diffusion instead of Midjourney?
- Pains of using Stable Diffusion
- Install Stable Diffusion WebUIs and manage models using Stability Matrix
- Step 1: Download and Install Stability Matrix.
- Step 2: Install Packages (WebUI)
- Step 3: Install Base Models
- Optional: Install extra checkpoints
- Step 4: Launch Stable Diffusion WebUI
- In Conclusion
What is Stable Diffusion?
Sable Diffusion is a text-to-image model released in 2022 by Stablity AI. The model has undergone continuous improvement, Stable Diffusion XL (SDXL) 1.0, the most recent model, added native 1024x1024 resolution and improved handling of human limbs and faces.
Why use Stable Diffusion instead of Midjourney?
Midjourney is an excellent tool for generating cinematic scenes and high-quality fashion photos. However, it’s not open-sourced; it has censorship, so there is no NSFW content. And all your data—images you upload as references and prompts—will be fed to Midjourney.
Stable Diffusion, on the other hand, is open-sourced, so you can generate whatever you like without being censored. If you are an artist and want to train a model based on your style, you can use Stable Diffusion as well.
Even better, Stable Diffusion is free!!!!!
Pains of using Stable Diffusion
Then why do many artists begin with Midjourney, DALLE, Leonardo AI, etc. instead of Stable Diffusion?
The main reason is that those AI art tools are web-based and have good UIs. Simply go to their website and you can generate AI art without downloading or installing anything on your computer. While Stable Diffusion is a bit complicated to use:
- Stable Diffusion requires a powerful graphics card to run locally, such as the RTX 4090 (probably the best option) I run Stable Diffusion on Macbook Pro M2, and it works, but the image generation is slow.
- You also need to install a web UI in order to run Stable Diffusion on your computer. However, there are too many web UIs—Automatic 1111, ComfyUI, Forge, Fooocus, etc.—and they support different features, so you might have to install them all depending on what art style and effect you want to create.
- The installation process is a bit complicated; you must install git, Python, homebrew, and PyTorch before installing web UIs 😫
- Next, you go to Huggingface and CivitAI, manually download the models, and place them in the WebUI's models folder before you begin to generate images. You might install more fine-tuning models for better results. Keep in mind that those models are between 2GB and 6GB, so make sure you have enough storage capacity. The good thing is that you can share the dependencies and models between web UIs.
- Finally, you can launch the Stable Diffusion Web UI. Open Terminal and run command like ./webui.sh in the web UI folder, and you should see your web UI is running at http://127.0.0.1:7860/
Sounds painful? I got you.
While I can’t persuade you to get an RTX 4090 😆, allow me to introduce "Stability Matrix" an incredibly helpful tool that takes care of the rest and lets you install everything with just a few clicks!
Install Stable Diffusion WebUIs and manage models using Stability Matrix
Stability Matrix is a WebUI and model management tool developed by LykosAI, allowing you to run Stable Diffusion without going through the painful manual installation process.
Let’s follow this step-by-step to use Stable Diffusion and generate images.
Step 1: Download and Install Stability Matrix.
Go to Stability Matrix Github page, click download, and install.
Step 2: Install Packages (WebUI)
For beginners, I recommend installing the original Stable Diffusion WebUI by Automatic 1111.
If you have been using Stable Diffusion for a while and want to build a more advanced workflow, I recommend installing ComfyUI.
Step 3: Install Base Models
We still need base models before generating images. Make sure you have enough storage space before importing models.
Click Model Browser > Hugging Face to install preferred base models. Here, I install Stable Diffusion 1.5 and Stable Diffusion XL. Check the models and click Import.
Optional: Install extra checkpoints
For creating certain art styles, you can further install related fin-tune checkpoint models.
Click Model Browser > CivitAI and search for models you like, then click Import.
Step 4: Launch Stable Diffusion WebUI
Now click Launch, select a WebUI in the drop-down menu, and then hit the green Launch button to run stable diffusion!
A new window will pop up and now you can generate AI art with Stable Diffusion!!
In Conclusion
Installing and running Stable Diffusion on your computer could be a painful process, especially if you want to use multiple Web UIs.
With Stability Matrix, you can install and manage all packages and models with ease.
