Cloud Stable Diffusion 3 and SwarmUI Tutorial

In this comprehensive article, we will explore how to use SwarmUI, Stable Diffusion 3, and other Stable Diffusion models on various cloud computing platforms. This guide covers three main options: Massed Compute, RunPod, and Kaggle. Each platform offers unique advantages for running these powerful AI image generation tools.

1. Using SwarmUI on Massed Compute

Massed Compute is one as the cheapest and most powerful cloud server provider, offering pre-installed SwarmUI and the latest version of Stable Diffusion models.

1.1 Registration and Deployment

To get started with Massed Compute:

Register Massed Compute as any website.

After registration, enter your billing information and load some balance.

Go to the "Deploy" section.

Select the template and deploy your virtual machine.

1.2 Connecting to the Virtual Machine

To connect to your Massed Compute virtual machine:

Download and install the ThinLinc client appropriate for your operating system.

Configure the ThinLinc client:

Open the ThinLinc client and click "Options."

Go to "Local Devices" and uncheck all options except "Drives."

Add a folder for synchronization to upload or download files.

Use the provided login IP address and password to connect to your virtual machine and start using like in your own Windows PC.

1.3 Using SwarmUI on Massed Compute

Once connected to your Massed Compute virtual machine:

Update SwarmUI to the latest version with 1-click updater button.

Access the SwarmUI interface through the virtual machine's desktop.

1.4 Generating Images with SwarmUI

To generate images using SwarmUI on Massed Compute:

Select your desired Stable Diffusion model (e.g., Stable Diffusion 3, SDXL, etc.).

Choose your preferred sampler and scheduler settings.

Enter your prompt and generate images.

Monitor GPU usage and generation speed using the provided tools.

1.5 Downloading Generated Images

To download your generated images from Massed Compute:

Navigate to the output folder within the SwarmUI directory.

Copy the output folder to your synchronization folder.

Access the synchronized images on your local machine.

1.6 Using CivitAI API for Model Downloads

A new feature allows you to download gated CivitAI models:

Obtain your CivitAI API key from your account settings.

Add the API key to your user info in SwarmUI.

Use the model downloader in the utilities section to access CivitAI models.

2. Using SwarmUI on RunPod

RunPod is another cloud service provider that offers powerful GPUs for running SwarmUI and Stable Diffusion models.

2.1 Registration and Deployment

To get started with RunPod:

Register RunPod as any other website.

Set up billing and load credits.

Go to the "Pods" section and click "Deploy Pod."

2.2 Selecting a Pod Configuration

When selecting a pod configuration:

Choose "Community Cloud" for cheaper usage.

Select "Extreme Speed" from the filters.

Choose NVME storage for better performance.

Select the desired RAM and GPU configuration (e.g., 3x RTX 4090).

Use the "RunPod PyTorch 2.1 with CUDA 11.8" template for best compatibility.

2.3 Installing SwarmUI on RunPod

To install SwarmUI on your RunPod instance:

Connect to JupyterLab.

Upload the installer file (install_linux.sh) to JupyterLab.

Use the terminal to run the installation commands.

Follow the on-screen instructions to complete the installation.

2.4 Using SwarmUI on RunPod

After installation:

Restart the pod once to ensure proper functionality.

Use the provided commands to start SwarmUI.

Connect to the SwarmUI interface using the HTTP port.

2.5 Generating Images and Managing Models

To generate images and manage models on RunPod:

Use the model downloader in the utilities section to add new models.

Configure multiple backends to utilize all available GPUs.

Generate images using your desired settings and prompts.

2.6 Downloading Generated Images

To download generated images from RunPod:

Use the "Download as archive" option for the output folder.

Alternatively, upload images to Hugging Face or use RunPodCTL for faster downloads.

3. Using SwarmUI on Kaggle (Free Account)

Kaggle offers a free option to use SwarmUI and Stable Diffusion models with some limitations.

3.1 Setting Up Kaggle Notebook

To set up SwarmUI on Kaggle:

Register for a free Kaggle account and verify your phone number.

Create a new notebook and import the provided Kaggle notebook file.

Select GPU T4 x2 as your accelerator.

3.2 Installing SwarmUI on Kaggle

Follow these steps to install SwarmUI on Kaggle:

Execute the provided cells in the notebook to download models and set up SwarmUI.

Follow the on-screen instructions to complete the installation process.

3.3 Configuring SwarmUI on Kaggle

After installation:

Modify the model root path to use the Kaggle temporary disk space.

Add additional backends to utilize both T4 GPUs.

Restart SwarmUI using the provided commands.

3.4 Generating Images on Kaggle

To generate images using SwarmUI on Kaggle:

Select your desired model and settings.

Generate images using both T4 GPUs for parallel processing.

Be aware of potential RAM limitations when using certain models or configurations.

3.5 Downloading Generated Images from Kaggle

To download your generated images from Kaggle:

Use the provided cell to zip all generated images.

Download the zip file from the Kaggle working directory.

Conclusion

This comprehensive guide provides detailed instructions on how to use SwarmUI, Stable Diffusion 3, and other Stable Diffusion models on Massed Compute, RunPod, and Kaggle. Each platform offers unique advantages and considerations:

Massed Compute provides a pre-installed, cost-effective solution with powerful GPUs.

RunPod offers flexibility and high-performance options for more advanced users.

Kaggle provides a free option with some limitations, ideal for those looking to experiment without cost.

By following this guide, users can choose the platform that best suits their needs and start generating high-quality AI images using the latest Stable Diffusion models and SwarmUI interface.