Generate a 3D model from a single image
Photogrammetry normally wants dozens of photos from every angle. This tool wants one. Upload a single image of an object and get back a 3D model you can rotate, light, and drop into a scene.
20 free credits at signup — enough for 5 runs of this tool. No card required.
- Cost
- 4 credits
- Input
- 1 upload
- Output
- 3D model
How 2D to 3D works
The model infers the geometry it can't see from the shapes it can, which makes it excellent for concepting and previews and less suited to work that needs dimensional accuracy. Objects photographed against an uncluttered background, fully in frame, and evenly lit produce the cleanest results. It's the most compute-heavy tool on the platform, which is why it's 4 credits.
Upload one image
One object, filling the frame, on a clean background.
Generate
The model reconstructs geometry, including the surfaces it can't see.
Download the mesh
Take it into your 3D tool of choice to refine, retopologise, or render.
What people use it for
- Turn a product photo into a rotatable 3D preview for a store page
- Block out game props and set dressing from reference images
- Build AR previews without a scanning rig or a modelling pass
- Give a 3D artist a starting mesh instead of a blank scene
2D to 3D questions
Is the model dimensionally accurate?
No. Geometry on the unseen side is inferred, not measured. It's built for concepting, previews, and starting meshes — not for engineering, fabrication, or anything requiring true measurements.
What kind of photo works best?
A single object, filling most of the frame, evenly lit, on an uncluttered background. Cluttered scenes and partially occluded objects give the model less to work from.
Why does it cost more than the other tools?
3D reconstruction is substantially more compute-intensive than a 2D pass. At 4 credits it's the most expensive tool on the platform, alongside Outpaint and Comic Panels.