Software Options for Stable Diffusion

When it comes to utilizing Stable Diffusion productively, you essentially have three popular options:

1. Automatic1111: This tool stands out for its user-friendly web UI and seamless integration with essential tools such as ControlNet, face reconstruction, and scaling. Additionally, it supports a wide range of features through extensions, including 3D image and video generation.
2. ComfyUI: While offering functionalities similar to Automatic1111, ComfyUI distinguishes itself with a unique node-based interface. This provides advanced users with the flexibility to customize their image generation pipeline. However, this interface might be somewhat challenging for beginners.
3. InvokeAI: Offers beautiful UI and supports additional functionalities that are not in Stable Diffusion itself such as scaling and ControlNet.

There are other alternatives available. Among other alternatives, some tools do not offer ease of use and seamless integration with key external packages such as ControlNet and GFPGAN. For those not looking to manually integrate these tools, I would hesitate to recommend them. I’ve included a brief list of these other options in the nextsection for those who are interested.

For beginners or those new to Stable Diffusion, I suggest starting with Automatic1111 for the wide variety of extensions to enhance its capability. InvokeAI UI is very nice, but support for a cool feature like AnimateDiff’s not there yet (https://github.com/invoke-ai/InvokeAI/issues/4477). To help you make a decision, I’ve created guides: Getting started with Automatic1111, Getting started with ComfyUI, Getting started with InvokeAI. These resources will provide further insights and assist you in making an informed choice.

Here are some additional choices for exploring Stable Diffusion:

1. Official Stable Diffusion Code on GitHub:
   • The official source code for Stable Diffusion is available at CompVis/stable-diffusion on GitHub. This resource is excellent for those who want to delve into the source code’s structure and understand its inner workings. I sometimes use it myself to modify code for both learning and experimentation. However, it’s important to note that this isn’t an end-user package. As a core module, its focus is not on user-friendliness or inclusion of other packages.
2. HuggingFace’s Diffusers Package:
   • For those comfortable with Python, HuggingFace’s Diffusers package offers another way to implement Stable Diffusion in your scripts. Detailed instructions can be found on HuggingFace’s blog. The package allows image generation with just a few lines of code. However, it’s not designed for end-users seeking a user-friendly interface or extensive external package integration.