Turn one T-pose picture into a rigged 3D character

Drop a single image of a character on the AutoRig home page. If it reads as a T-pose or A-pose with limbs clear of the body, you get a 3D model, a skeleton, retargeted animations and every export format back as one task.

Automatic pose detection
Model, rig and animations in one task
GLB, Unity FBX, HDRP package, Blender
Upload a picture

One picture in, a whole character out

Most image-to-3D tools stop at a mesh. You get a static model, and rigging it is still your evening. AutoRig treats the picture as the start of the whole pipeline: the image becomes a textured 3D model, the model gets a skeleton and skin weights, and an animation library is retargeted onto it. All of it arrives as a single task with a single progress page.

The routing decision is made once, right after upload. The site looks at your picture and decides whether it shows a character that can carry a skeleton. If it does, you get generation plus rigging. If it does not — a vehicle, a prop, a building, a character whose limbs are fused into the body — you still get a proper PBR 3D model, because the model is what you paid for either way.

Characters generated from a single image

Every preview below is a finished public task on this site. Each one started as one uploaded picture and ended as a rigged character with animations. Press play, then open the task page to inspect the result.

Cute Pink Creature with Large Eyes and Playful Expression

17 files delivered

Open the task

Fantasy Dwarf Adventurer with Green Tunic and Beard

17 files delivered

Open the task

Armored Knight with Spiked Shoulder Pads and Mask

9 files delivered

Open the task

Mystical Bear Shaman with Staff and Orbs

9 files delivered

Open the task

The clips are the preview videos the pipeline renders for every finished task. They show the generated character actually playing back retargeted animation, not a turntable of a static mesh.

How it works, step by step

Four stages run without you touching anything after the upload.

  1. Drop the picture

    Open the AutoRig home page and drop a PNG, JPG or WEBP file on the upload zone. One image is one character. You need to be signed in, because generation is charged to an account.

  2. The site decides what it is looking at

    Detection checks whether the image shows a character in a T-pose or A-pose with limbs clear of the torso. This decides the route only: riggable pictures go to generation plus rigging, everything else goes to generation alone.

  3. The 3D model is generated

    The picture goes to the GPU farm and comes back as a textured mesh with PBR materials, retopologised into usable geometry rather than a raw scan-like surface.

  4. The model is rigged and animated

    A skeleton is fitted, skin weights are solved, and the animation library is retargeted onto the new character. The task then publishes GLB, Unity FBX, the HDRP unitypackage, Blender files and a preview video.

What lands in the task when it finishes

A finished character task publishes a full delivery set, not a single file. The examples on this page shipped between 9 and 17 artifacts each.

Rigged GLB

The web-native format. It plays in the browser viewer on the task page and drops straight into three.js, Babylon, Godot or any glTF pipeline.

Unity FBX with animations

The skeletal mesh with the retargeted animation set baked in, ready for Mecanim.

HDRP unitypackage

A Unity package with materials wired for the High Definition Render Pipeline, so the character is not pink on import.

Blender files

The scene as Blender sees it, for weight painting, retopology touch-ups, or exporting to something the list above does not cover.

Preview video

A rendered clip of the character playing its animations - the same videos embedded on this page.

What makes a picture riggable

The detector is looking for a readable body, not for art quality. A rough sketch in a clean T-pose beats a beautiful render of a character folded into a dramatic action pose.

And when the picture is not a character

Nothing is rejected. A picture that cannot carry a skeleton still produces a textured PBR 3D model - you simply get the model without the skeleton and the animations. That is a routing decision made per picture, never a failure message.

Several pictures at once

Drop a handful of images and each one becomes its own task, queued together. The upload zone accepts a multi-file selection and starts them in sequence, so a set of concept sheets can turn into a set of rigged characters in one pass. Drop a single file and you land on that task's page; drop several and you land on your task list.

What it costs, and what is free

Free

Uploading a 3D model you already have and running the rig costs nothing and needs no account. Watching the result costs nothing either - the browser viewer, the animation playback, the preview video and the public gallery are all open. AutoRig is free to run and free to look at.

Paid

Two things spend credits. Generating a character from a picture costs 3 credits, charged when the task is created, because it books GPU time on the farm. Unlocking a finished task's downloads costs 10 credits and releases the whole delivery set: the ZIP, the protected files, animal variants and custom animation downloads. A single extra animation is 1 credit.

Credits are sold in packs: 30 for $3, 100 for $10, 500 for $50 and 1000 for $100 - a flat $0.10 per credit at every tier, with roughly 20% off when paying with crypto. At that rate one task unlock is $1 and one image-to-character generation is $0.30.

If the pipeline breaks rather than the picture, the 3 credits go back to your balance automatically.

One thing to be straight about: there is no automatic free credit allowance at the moment. New accounts start at zero, so the first generation or the first download needs a credit pack. Everything described as free above stays free without one.

Questions people actually ask

Do I need a 3D model to start?

No. That is the point of this page - a single picture is enough. If you already have a GLB, FBX or OBJ, upload it directly instead and skip generation entirely.

Does the character have to be in a T-pose?

A T-pose or an A-pose both work, as long as the limbs are clear of the body. The pose decides whether rigging is attempted; it never decides whether you get a model.

How long does it take?

The examples on this page took roughly 20 to 50 minutes end to end, including time spent waiting in the farm queue. The task page shows live progress, and you can close the tab - the task keeps running.

What if I upload something that is not a character?

You get a textured PBR 3D model of whatever it was. Detection routes the job, it does not reject it.

Can I rig several characters at once?

Yes. Select several images in the upload zone and each becomes its own task, charged separately.

Do I need an account?

For generating a character from a picture, yes - credits live on accounts. For uploading a 3D model you already have and running the rig, no.

Related guides

More on poses, formats, and what happens after the rig:

T-Pose vs A-Pose GLB Auto Rigging FBX Auto Rigging Animation Retargeting How It Works Gallery Buy Credits FAQ

Try it with one picture

Open the home page, drop an image of a character, and watch it come back rigged.

Upload a picture