Input photograph
Choose an architectural photograph from your device.
Computer vision · geometry · architecture
An experimental system using image analysis to reconstruct simplified architectural geometry from a photograph.
Concept visualization — not an output produced by the algorithm.
About the project / 01
Image → 3D explores how to extract a building’s main planes from a single image and turn them into lightweight geometry.
Rather than aiming to recreate every detail photorealistically, the project focuses on dominant surfaces and skips irregular elements that complicate simplified reconstruction. The process produces a 3D model that can be downloaded as a GLB file.
Explore the technologyA single photograph as input
↗Dominant architectural planes
⌗Lightweight geometry, not only a point cloud
◇Export to the GLB format
↓Workflow / 02
These four stages describe the intended workflow; analysis details and results depend on processing.
Choose an architectural photograph from your device.
The system searches for dominant planes and geometric features.
Recognized surfaces provide the basis for a simplified model.
When processing completes successfully, download the result as a GLB file.
Interactive tool / 03
Upload a photo, follow processing progress, and view the model when it is ready.
Choose a photo from your device