ComfyUI ControlNet Auxiliary Preprocessors 🧩 Workflow Open source

What it is

comfyui_controlnet_aux by Fannovel16 (Apache-2.0) bundles the preprocessors that turn a reference image or video frame into a control map: depth, OpenPose body and hand pose, Canny and HED edges, line art, normal maps, segmentation and more. Those maps guide ControlNet and similar models so generated images and videos follow a composition, pose or layout. For content work that means matching a product's exact placement, keeping a character's pose across frames, or turning a sketch into a finished illustration. Most shared ControlNet workflows list this pack as a requirement.

Key features

• Depth estimators (Depth Anything, MiDaS, Zoe and others)
• OpenPose and DWPose body, hand and face keypoints
• Canny, HED, line art, anime line art and scribble extractors
• Normal maps, segmentation and tile preprocessors
• Models auto-download on first use
• Batch-friendly for video frames

How to install

Install via ComfyUI-Manager (recommended), run the bundled install.bat on portable Windows builds, or git clone https://github.com/Fannovel16/comfyui_controlnet_aux/ into custom_nodes and pip install -r requirements.txt.

Pricing and rights

Free, Apache-2.0. Individual preprocessor models carry their own licences, some non-commercial, so check the model card before using outputs in client or commercial work.

Who it is for

Illustrators, product and fashion creators, and video makers who need to control pose, layout or depth in generations rather than relying on prompts alone.

Verdict

A standard building block for controllable generation, actively maintained and used by most ControlNet tutorials. Dependency installs (for example ONNX runtimes) can be fiddly on some systems.