Image to OpenSCAD

You upload a photo and optional size and notes. A vision AI model writes OpenSCAD code for the object, with important measurements as variables. OpenSCAD compiles the code in your browser, and if it fails, the errors are sent back to the AI to fix, up to two times.

The examples are free to explore without an account. To use your own photo, sign in: free accounts get 3 photo tries and can rotate and inspect every result in 3D. Premium adds STL and .scad downloads, editing in the OpenSCAD playground and 30 photos a month.

A photo has no scale, so the AI estimates sizes unless you give them. Enter the overall width, depth and height and the model matches them. Small details are approximations; check critical measurements before printing.

Mechanical and geometric parts such as brackets, hooks, clips, stands, boxes and knobs. Organic shapes such as people, animals and sculptures are reduced to simple blocks, because OpenSCAD builds models from primitives.

The photo is sent to the AI model only when you click Generate, and only to create the model. It is resized in your browser first. The OpenSCAD compile runs locally in your browser tab.

Yes, with Premium. The result is plain OpenSCAD code with named parameters. Open it in the OpenSCAD playground to change the values, edit the code, or ask the playground's AI chat for changes, then export STL or 3MF.