Image → 3D
Early version
Photo → 3D Mesh
Lift a photo into a textured relief mesh you can export and model with.
The problem
Point clouds are hard to model with — most students need an actual surface they can drop into Rhino, Blender, or SketchUp. Rebuilding a reference as geometry by hand is slow.
What this does
Same on-device depth estimation as the point-cloud tool, but the depth map is turned into a displaced grid mesh with the photo as its texture. Export to OBJ or glTF and bring it straight into your modelling software as a relief / base surface.
Try it
Loading tool…
What it does
The estimated depth map is used as a heightfield: a regular grid is displaced along its normal by the depth value at each vertex, and the original photo is applied as a texture. The result is a continuous surface (a "relief" of the image) rather than loose points — ready to export.
Why it's useful
- Get a modellable surface from a single reference image.
- Export OBJ or glTF/GLB and open it in Rhino, Blender, SketchUp, or Unreal.
- Useful for façade studies, terrain-like reliefs, and quick massing references.
Honest limits (early version)
- It's a 2.5D relief, not a watertight solid — there's only one viewpoint of depth.
- Steep depth discontinuities stretch the mesh; a depth threshold trims the worst of it.
- Decimate / retopologise in your modeller for clean geometry.
The real pipeline (roadmap)
True single-image-to-mesh models (e.g. TripoSR, InstantMesh) hallucinate the unseen sides but need a GPU. This artifact is the honest, local, geometry-you-can-trust version.
Built on
- Depth Anything V2 (depth model)Apache-2.0 (small model)
- three.js (mesh + OBJ/GLTF export)MIT
- TripoSR (roadmap reference)MIT