So you’ve fired up Invoke AI on Docker and your browser’s whispering sweet prompts at you. But wait — where’s the magic? You need a model, my friend. And not just any model — let’s go straight for the big one: SDXL (Stable Diffusion XL), the extra large upgrade with smarter brains and prettier pixels.
Here’s how to download and load your first SDXL model into Invoke AI — all without breaking your Docker spellbook.
Step 1: Locate Your Models Folder
Remember the folder we created back in Part I?
C:\invokeai-docker\models
Inside there, you’ll want to create a new folder for your SDXL model:
mkdir C:\invokeai-docker\models\sdxl
Step 2: Download the SDXL Base Model
You’ll need the official Stable Diffusion XL 1.0 Base model from Hugging Face. (Yes, you need a free account there.)
Go to: https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0
Click Files and versions.
Download:
sdxl_base_1.0.safetensors
Put file into:
C:\invokeai-docker\models\sdxl
Step 3: Download the SDXL Refiner
For next-level detail, let’s also download the SDXL Refiner, go to:
https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-1.0
Download:
sdxl_refiner_1.0.safetensors
Place it in the same models\sdxl folder.
Step 4: Go to the Web UI
Open your browser and navigate to:
http://localhost:9090
Step 5: Click on the 3D Cube on the Left Sidebar
Click on the “Models” icon on the left, it’s shaped like a 3D Cube, to open the Model Manager and import the models.
Step 6: Click on “Scan Folder” tab on the right side
Under Add Model, click on the Scan Folders tab and type under folder:
/invokeai/models
Step 7: Click the “Scan Folder” Button Then Click Install All
Click the Scan Folder button, then click Install All. The stable-diffusion-xl-base-1.0 and stable-diffusion-xl-refiner-1.0 models will be installed. This may take a few minutes, when it’s complete it will look like this:
Step 8: Click on the Square with little squares on the corners on the Left Sidebar
Click on the “Canvas” icon on the left, it’s shaped like a 2D square with little squares on each corner. This will bring you back to the main page. Click the Model dropdown in the top menu.
Step 9: Type Something in the Positive Prompt
Type a image prompt in the the Positive Prompt box, for example, “Cute animals in a calm meadow.”
Look for sdxl_base_1.0 under Model — load it.
Step 10: Click the Invoke Button
Boom — you’ve generated your first image using SDXL locally, powered by your GPU.
