AUTOMATIC1111 (often shortened to A1111) and ComfyUI are two popular open source web UIs for Stable Diffusion. Both allow you to interactively develop image generation pipelines, so which one should you use?
| A1111 | ComfyUI |
|---|---|
| Beginner-friendly | Steeper learning curve |
| Built-in common workflows (e.g. text-to-image, image-to-image) | More flexibility in customizing workflows |
| Fewer cutting-edge features | Strong momentum and pace of development |
Takeaways
- If you are a beginner and just want to quickly generate AI art: Use A1111
- If you are somewhat familiar with Stable Diffusion and want to run more complex image generation pipelines: Use ComfyUI
What is A1111?
A1111 is one of the first open-source web UIs created for Stable Diffusion. The project is technically called “Stable Diffusion web UI” and the author’s handle is AUTOMATIC1111. However, over time people started calling the project by the author’s handle or just A1111 for short.
A1111 was created in August 2022, coinciding with the first public release of Stable Diffusion. It quickly became one of the most popular user interfaces for running Stable Diffusion locally.
Built using Gradio, the UI feels immediately familiar, and this is its greatest strength; it’s very easy to jump in and quickly generate AI art.
Inpainting (i.e. editing specific areas of an image) is a good example of a workflow that A1111 makes easy:
Generate the image in the
txt2imgtab
Click a button to send the generated image to the
img2imginpaint tab
Brush over the area you’d like to draw over (a.k.a. the mask)
Update your prompt
Though A1111 is very user-friendly, many complain about its reliability. It’s also not exactly on the cutting edge; as of this writing it still does not support Flux. Many users have started switching over to newer frameworks such as Forge, Fooocus, or ComfyUI.
What is ComfyUI?
ComfyUI has become one of the fastest growing open-source web UIs for Stable Diffusion. ComfyUI was created in Jan 2023 and has positioned itself as a more powerful and flexible version of A1111.
In ComfyUI, you define workflows down to the node level in a flowchart-like UI. The ability to tweak node-level parameters and also write custom nodes gives you more flexibility compared to A1111.
Here are some particularly impressive ComfyUI workflow examples:
ComfyUI’s community and pace of development is also growing faster. Most recently, the author of ComfyUI, comfyanonymous, founded Comfy Org, a team dedicated to improving the reliability of core ComfyUI. They’ve already released:
- comfy-cli, a handy tool for easily installing ComfyUI, managing custom nodes, and running workflows programmatically
- Comfy Registry, a custom node “app store” to standardize and improve the reliability of custom node development
Below is a simple workflow for text to image using Flux on ComfyUI:
You can run this workflow on Modal with this gist.
Conclusion
A1111 is great for beginners who want to quickly generate AI art using fairly straightforward workflows like text-to-image or image-to-image.
While ComfyUI has a steeper learning curve, more advanced users will be able to design powerful workflows, especially when leveraging the rich custom node ecosystem. Because of its strong momentum and pace of development, it’s likely also more future-proof.
A1111 vs ComfyUI doesn’t necessarily have to mean one vs. the other. Many users use A1111 for rapid prototyping and also ComfyUI for serving complex workflows in production.
Running A111 and ComfyUI on Modal
It’s easy to try out both on Modal’s serverless infrastructure:
By running image generation on Modal, you can:
- Avoid the headache of locally installing these pieces of software
- Only pay for the GPU time you use
- Host your workflow behind a production-ready API endpoint that autoscales to handle any traffic load
