ComfyUI
by comfyanonymousNode-based UI for Stable Diffusion. Wire up complex pipelines with full control — the power-user choice.
About ComfyUI
ComfyUI is a Python open-source image & vision ai project by comfyanonymous. Node-based UI for Stable Diffusion. Wire up complex pipelines with full control — the power-user choice. With 60,000 GitHub stars and 0forks, it's one of the Image & Vision AI tools worth knowing. You can download the source, browse it on GitHub, or read the full setup guide in the README below.
ComfyUI — guide
The pipeline is the interface
ComfyUI does not hide how an image gets made. Loading a checkpoint, encoding a prompt, sampling, decoding the latent — each is a node on a canvas, and you connect them yourself. The first hour is genuinely harder than any other Stable Diffusion front-end. What you get in return is that nothing is happening off-screen: when a result is wrong you can see which stage produced it, change that one node, and rerun.
Why people put up with the learning curve
Workflows are files. A graph that produces a particular look can be saved, version-controlled, handed to someone else, or dropped back in six months later and reproduced exactly. ComfyUI also embeds the workflow inside the PNGs it generates, so dragging an image back into the canvas rebuilds the graph that made it. For anyone running the same treatment across hundreds of images, that reproducibility is the feature.
What it costs to run
A Python install and a GPU — 8GB of VRAM is a comfortable floor, less will work with smaller models. Models are not bundled; you download checkpoints, LoRAs and VAEs yourself and drop them into the right folders, which is the step that trips most people up first. Everything runs locally, so nothing you generate leaves your machine.
When to pick something else
If you mainly want to make pictures rather than build pipelines, InvokeAI gives you a paintable canvas and keeps the graph out of sight. Reach for ComfyUI when you need control over the process itself, not just the output.
From the project README
<div align="center">
ComfyUI
**The most powerful and modular AI engine for content creation.**[![Website][website-shield]][website-url] [![Dynamic JSON Badge][discord-shield]][discord-url] [![Twitter][twitter-shield]][twitter-url] [![Matrix][matrix-shield]][matrix-url] <br> [![][github-release-shield]][github-release-link] [![][github-release-date-shield]][github-release-link] [![][github-downloads-shield]][github-downloads-link] [![][github-downloads-latest-shield]][github-downloads-link]
[matrix-shield]: https://img.shields.io/badge/Matrix-000000?style=flat&logo=matrix&logoColor=white [matrix-url]: https://app.element.io/#/room/%23comfyui_space%3Amatrix.org [website-shield]: https://img.shields.io/badge/ComfyOrg-4285F4?style=flat [website-url]: https://www.comfy.org/
[discord-shield]: https://img.shields.io/badge/dynamic/json?url=https%3A%2F%2Fdiscord.com%2Fapi%2Finvites%2Fcomfyorg%3Fwith_counts%3Dtrue&query=%24.approximate_member_count&logo=discord&logoColor=white&label=Discord&color=green&suffix=%20total [discord-url]: https://discord.com/invite/comfyorg [twitter-shield]: https://img.shields.io/twitter/follow/ComfyUI [twitter-url]: https://x.com/ComfyUI
[github-release-shield]: https://img.shields.io/github/v/release/comfyanonymous/ComfyUI?style=flat&sort=semver [github-release-link]: https://github.com/comfyanonymous/ComfyUI/releases [github-release-date-shield]: https://img.shields.io/github/release-date/comfyanonymous/ComfyUI?style=flat
This is an excerpt from the project's own README, quoted here for reference. Read the full README on GitHub →
ComfyUI — FAQ
Is ComfyUI hard to learn?
Harder than prompt-box front-ends, yes — you assemble the generation pipeline node by node. The usual route is to start from a working template workflow, change one node at a time, and build understanding from there rather than starting on an empty canvas.
Can I share a ComfyUI workflow with someone else?
Yes, and this is a genuine strength. Workflows save as files, and generated PNGs carry the workflow embedded in them — dragging such an image onto the canvas rebuilds the exact graph that produced it.
What GPU do I need for ComfyUI?
8GB of VRAM is a comfortable starting point. It runs on less and on Apple Silicon, but larger checkpoints and higher resolutions will either be slow or fail to fit in memory.
Does ComfyUI include the models?
No. You download checkpoints, LoRAs and VAEs separately and place them in the matching folders under the models directory. Missing or misplaced model files are the most common reason a first workflow fails to run.
More Image & Vision AI skills
See all Image & Vision AI →The de-facto web UI for running Stable Diffusion locally. Plugins, LoRAs, ControlNets, img2img, in-painting — it all started here.
Meta's foundation model for image segmentation. Click any pixel, get a perfect mask of the object it belongs to.
Professional creative tooling for Stable Diffusion. Unified canvas, layers, workflow tabs — designed for actual production use.