PhotoMesh — automated photogrammetry and 3D reconstruction

Overview

PhotoMesh is a fully automated photogrammetry engine: it turns a set of photos (drone, ground camera, or a mix) into a photo-realistic 3D model of a site or asset. Computer-vision and computational-geometry algorithms find the connections between photos, reconstruct geometry and apply texturing automatically — no manual stitching.

Two scales of work

• 3D mapping — city-, pit- or infrastructure-scale models with survey-grade georeferencing, for urban planning, infrastructure management and 3D GIS;
• Digital twin for inspection (close-range) — models accurate to fractions of a millimeter for remote condition monitoring of structures, removing the risk of human error and the cost of traditional on-site inspection.

Output exports to a wide range of formats — 3D mesh, point cloud, orthophoto, digital surface and terrain models (DSM/DTM) — for use in TerraExplorer, third-party GIS, or a company's own ML/AI analysis tools.

Performance and scale

PhotoMesh is built for projects of any size, from a single asset to hundreds of thousands of photos. Its tiled architecture and distributed-processing support (across multiple fuser machines or the cloud) let it process tens of square kilometers a day, and the REST API allows it to be wired into an automated production pipeline with no operator intervention.

For industrial sites in Kazakhstan, PhotoMesh covers a job that used to require costly field survey or laser scanning: regular, repeatable 3D monitoring of pits, dumps, sites and linear assets from ordinary drone imagery.