Online OpenPose Editor for ControlNet
Pose a 3D mannequin and export the matching OpenPose skeleton (COCO-18 keypoints, standard colours) as a PNG or JSON for ControlNet. Free, no sign-up.
Opens in OpenPose skeleton mode · free, runs in your browser
How it works
Pose the figure
Drag joints in 3D, start from a library pose or import a photo. Joint limits keep the pose anatomically possible.
Frame the camera
Pick the angle and focal length your final image needs. The skeleton is projected from exactly that camera.
Export for ControlNet
Download the black-background skeleton PNG or the OpenPose JSON and load it into a ControlNet OpenPose unit.
Why edit OpenPose in 3D?
Flat OpenPose editors move dots in 2D, so limbs stretch, shrink and bend the wrong way as soon as you change the angle. Posing a jointed 3D figure keeps bone lengths fixed and joints within their natural range, and the skeleton is projected from a real camera. Foreshortened arms, low-angle hero shots and turned heads come out right without guessing where each keypoint should go.
Compatible with ControlNet and ComfyUI
The export uses the 18-point COCO body layout and the standard limb colours of the ControlNet OpenPose preprocessor, so you can feed the PNG straight into an OpenPose ControlNet with the preprocessor set to none. The JSON follows the OpenPose people/pose_keypoints_2d format that most OpenPose editors and ComfyUI nodes import.
Start from a photo or a pose library
You do not have to start from a T-pose. Open any pose from the library, or upload a photo and let Photo to Pose detect the body in your browser, then fix the details on the mannequin before exporting. Your photo is never uploaded to a server.
Frequently asked questions
Is the OpenPose editor free?
Yes. Posing, OpenPose PNG and JSON export are free and need no account.
Which keypoint format does the export use?
COCO-18 body keypoints (nose, neck, shoulders, elbows, wrists, hips, knees, ankles, eyes and ears) in the order and colours used by the ControlNet OpenPose preprocessor.
Which ControlNet preprocessor should I choose?
Choose none (or no preprocessing), because the image is already a skeleton, and pick an OpenPose ControlNet model that matches your base model.
Are my photos uploaded when I start from a photo?
No. Photo to Pose detects the pose inside your browser, so the image never leaves your device.