Custom models

Odyssey supports using custom Stable Diffusion models with one key distinction - models must be converted to CoreML - which is Apple's framework for supporting AI models locally.

There are two primary ways to find models that Odyssey supports. First, visit the CoreML community on Huggingface. There is a large list of models available to download and the repository is constantly updated. Second, for more advanced users, you can convert PyTorch models yourself by following this tutorial.

When you download a model, you'll need to take the following steps to use your model in Odyssey:

• Unzip the file

• In Odyssey select settings (⌘ ,)

• Navigate to Custom models

• Click the + sign

You'll now need to set the model up correctly to run with Odyssey.

• First, name your model. This will be the name that appears in your dropdown.

• Next, select the location where your model's folder is. You should select the folder that contains more than just the model (meaning the .txt file and other files.)

• Choose the base model your custom model was built off of. This should be outlined in the HuggingFace repository. Your options are Stable Diffusion 1.5, Stable Diffusion 1.5 (ControlNet), Stable Diffusion 2.1, and Stable Diffusion XL. To know whether your not your model has ControlNet support, look for _cn in the file name

• Select whether or not your model supports ControlNet or Safe Mode

• Import your model

Once your model is imported, you should be able to select your model from the model dropdown when running the Stable Diffusion node locally.

You can easily swap between models via this dropdown.

If you have any questions or need any assistance with custom model support, please reach out to help@odysseyapp.io or join our Discord server.