Skip to main content

Stable Diffusion 3 Installation

Stability AI has just released Stable Diffusion 3, which is a game-changing model that allows for much more photorealism, typography, and much more. This tutorial will guide you into how you can download and install Stable Diffusion 3 in your local environment. We will also try several worthy prompts to review how well is the new Stable Diffusion 3.

 

Stable Diffusion 3 Installation
Stable Diffusion 3 Installation

Step 1: Accessing the Official Website

Accessing the Official Website
Accessing the Official Website

First, let’s use any browser you are comfortable with and type ‘’Stable Diffusion 3’’ on the search bar. Go to the official Stability AI website where one can see all the features and novelties the Stable Diffusion 3 has to offer. On the site, you can view the images that are available and there are the lists of sample images and the tab with the button ‘Download Now’.

Step 2: Downloading from Hugging Face

https://huggingface.co/stabilityai/stable-diffusion-3-medium

Downloading from Hugging Face
Downloading from Hugging Face

Click on the “Download Now” button, which will bring you to the Hugging Face repository of Stable Diffusion 3. All the sample images and all the descriptions of the models to be estimated can be placed here as well. Before one can download anything from the Hugging Face repository, one is required to login or create an account. If you already have an ID, you can login, else click on the Word “Sign up”.

Step 3: Signing Up for Hugging Face

Signing Up for Hugging Face
Signing Up for Hugging Face

Enter your email address, password and choose your username for the site. Finally click the button marked create account. You will be asked to give purer details like name, e-mail, country, organization, preferred notification type, and more. Complete the details and press on “Agree and Access Repository Button”.

Step 4: Confirming Your Email Address

You will also get an email with this update in case you provide your email address. After that, on the same page, to complete the registration process, click on the ‘Confirm Email’ button. After that, your Stable Diffusion 3 Hugging Face repository will be unlocked, and you can then proceed to follow the instructions in step 5.

Confirming Your Email Address
Confirming Your Email Address

Step 5: Downloading Comfy UI

Downloading Comfy UI
Downloading Comfy UI

On the website of Hugging Face repository, there is the click tab that when selected will open the files and versions option In particular is the comfy UI. Click on it and you have to look down for a while to find the option to install Comfy UI to the system you are using Windows. Please, go to the direct link and download the 1. 4 GB file.

Step 6: Installing Comfy UI

After downloading, uncompress the file into a new folder which was created with the name “SD3”. By opening the folder, you get a few more folders and out of these “comfy_ui” is one of them. Open the “comfy_ui” folder and here you have opportunities to keep your model safe, nets and many others.

Step 7: Running Comfy UI

You’ll find two options: These make up the commands for a CPU to run, and commands for Nvidia GPU to run as well. If you have Nvidia click the icon ‘Run CPU’ or If you have AMD click the icon ‘Run AMD’. This will launch what is called Comfy UI which allows one to load models, input prompts, and do basic manipulations with parameters.

 

Running Comfy UI
Running Comfy UI

Step 8: Downloading and Exporting Models

Return to the Hugging Face repository and download: the Stable Diffusion 3 model (4.34 GB) and some test encoders. Once downloaded, copy and paste the model into the comfy – UI checkpoints folder and the test encoders into the comfy – UI models and clip folders.

Step 9: Testing Prompts

If you open Comfy UI there will be an interface where you can see the installed model at the bottom of the screen. Okay, now we try some prompts. You can use the demo prompts available on the Hugging Face repository site here . We tested two prompts: “A colorful street art wall splitter that reads B medium SD3” and a picture of a yellow man dressed in black. The results were quite remarkable, especially since the model gave rather high-quality images within a couple of seconds.

 

0
    0
    Your Cart
    Your cart is emptyReturn to Courses