●
3DAI

Auto Rig a 3D Model with AI

Upload a humanoid character standing upright and AI adds a skeleton and skin weights, so it can bend and move. Test the rig right here, then download a rigged GLB for Blender, Unity, Unreal or Godot.

Animal, bird, snake or another shape? Rig any shape free in your browser.

Loading the rigging tool...

Rig any shape free, in your browser

Pick a body type, tap a few points on your model (chin, wrists, paws, tail tip...), and get a skeleton with skin weights you can pose and download as a GLB. It works for people, four-legged animals, birds, snakes, tentacles and your own custom shapes. It is free, needs no sign-in, and runs on your device: your model is not uploaded.

Limits: parts that touch or are fused together (a long coat, folded wings, legs pressed together) may stretch when they bend. For the best humanoid quality, use the AI rig above.

Loading the free rigger...

How to rig a 3D model with AI

  1. 1

    Upload your model

    Drop a GLB or OBJ of one humanoid character, standing in a T-pose or A-pose. Big models are reduced to 50,000 faces in your browser.

  2. 2

    AI rigs it

    UniRig places the bones and paints how strongly each bone moves each part of the mesh. It usually takes 1 to 3 minutes.

  3. 3

    Test the pose

    Raise the arms, turn the head and bend the spine with the sliders. The skeleton is drawn on top so you can see every bone.

  4. 4

    Download

    Get the rigged GLB and animate it in Blender, Unity, Unreal or Godot, or retarget an animation onto it.

No model yet?

Make a character from a picture with image to 3D, then bring it here to rig it. Ask for a T-pose in the picture for the best rig.

Want new colours first?

Repaint your model with AI 3D texturing, then rig the textured model.

Frequently Asked Questions

Rigging gives a 3D model a skeleton: bones joined in a hierarchy, plus skin weights that say how much each bone moves each vertex. Without a rig a model is a statue. Auto rigging lets AI do this in one step: UniRig looks at the shape, places the bones and computes the weights, so you can pose and animate the model.
GLB and OBJ, with one character per file (have an FBX? Export it as GLB from Blender first). The AI takes up to 50,000 faces and 25 MB. A bigger model is reduced in your browser before it is sent, or you can shrink it first with the 3D model compressor. The result is always a GLB, which Blender, Unity, Unreal and Godot open.
A T-pose or A-pose works best: standing, arms held away from the body, legs apart, hands not touching anything. Parts that touch get rigged together. In the picture above the hands are in the pockets, so raising the arms drags the hips and jacket with them.
The AI rig is built for humanoid, standing characters: a head, two arms and two legs, upright. After each rig we check that the skeleton stands upright and fills the character; when it does not, you get your credits back and no rig. For animals, birds, snakes, tentacles and other shapes, use the free rigger lower on this page: you tap a few points and it builds the skeleton in your browser. Cars and rigid props usually need hand-made joints in a 3D app.
No. You get the skeleton and skin weights only, with no animation. The bones have generic names (bone_0, bone_1 and so on), not Mixamo or Unreal names. To use a ready-made animation, rename or retarget the bones in Blender, or use a retargeting tool that maps one skeleton onto another.
Each rig costs 30 credits. A 30,000-face character took about one minute in our test; if the AI has to start up first it can take a few minutes more. If the AI fails, takes longer than 15 minutes, or the skeleton does not fit a standing character, the credits are refunded. Files that are refused before rigging cost nothing.
Your model is sent to our server and to UniRig, an open-source rigging model by VAST-AI-Research (MIT licence), which runs on Replicate. Our copy is deleted when the rig is done. The rigged GLB link expires after about an hour, so download it once you are happy with the pose test.

Powered by UniRig (MIT, VAST-AI-Research).