Before anything else, our GNSS engine post-processes PPK kinematic and static captures against a PPP solution of the base-station files—delivering accurate, consistent base positions and removing errors in raw base-station data. It runs ahead of VPS and 3D reconstruction.

Every image runs through our ML models for segmentation, object detection, and scene understanding. Faces, people, and vehicles are automatically blurred for privacy—and you can pull additional object classes from our model library, plus temperature where captured.

Our VPS positions imagery at accurate metric scale—no LiDAR. A single capture is large—≈10,000 shots / 30,000 images, up to ~60 minutes—covering big spaces in one run, with cameras at any height (even upside down) for 2D baseline or full 3D. Fusion VPS then combines overlapping captures into large cells, so you can map spaces of virtually any size—massive coverage other platforms can't match. Driving stitches overlapping passes into one contiguous, edge-consistent surface by automatically matching neighbor images; indoors, big per-floor cells stack in 3D with accurate vertical alignment through stairwells.

Builds 3D meshes, point clouds, and ortho slices (ceiling- and floor-facing) from both capture cells and fusion cells. A smart geographic sharding system guarantees precise alignment at cell edges, and mesh quality is configurable by triangle size and texturing.