own/metal
← Image Generation tool · /COMFYU

ComfyUI

Node-based UI for Stable Diffusion, Flux and beyond.

// github

★ 115.5k

last commit · today

moderate GPU required GPL-3.0

// readme · what it is

ComfyUI is where the serious image-gen community lives. The node-graph workflow exposes every step of the diffusion pipeline — once you grok it, you can build pipelines no slider-based UI could express. Supports SD 1.5/SDXL/SD3/Flux, ControlNet, LoRAs, AnimateDiff, video models — basically the whole open-source frontier.

// deploy notes

SD 1.5 runs on 6GB VRAM. SDXL wants 12GB. Flux/SD3 want 24GB. Plan disk for model weights — 100GB fills fast.