✦
Summary
Flux ControlNet Collections is a repository of pre-trained ControlNet checkpoints for the FLUX.1-dev text-to-image model by Black Forest Labs, published by XLabs AI. ControlNet adds conditional control to image generation, letting users guide output with a reference image such as an edge map or depth map instead of relying on text alone.
Key Features
Pre-trained ControlNet checkpoints for the FLUX.1-dev model
Three models included: Canny, HED and Depth (Midas)
Models trained at 1024x1024 resolution for high-quality output
v3 versions usable directly in ComfyUI
ComfyUI custom nodes and ready-made workflows for each model
Python inference scripts and a Gradio demo for trying the models
Example generations showing conditioned results for each model
Released under the FLUX.1 [dev] Non-Commercial License
About Flux Controlnet Collections
The collection includes three models — Canny, HED and Depth (Midas) — each trained at 1024x1024 resolution, with v3 versions that can be used directly in ComfyUI. The repository provides example results, inference scripts, ComfyUI workflows and custom nodes, so users can try the models with a Gradio demo, a Python script or inside ComfyUI.
The collection is aimed at AI artists, researchers and developers who want precise control over FLUX image generation, for example producing cinematic photos or full-HD images conditioned on a specific structure. The weights are released under the FLUX.1 [dev] Non-Commercial License.