Laser Scanning & Reality Capture: The Complete Guide

Laser scanning (LiDAR-based reality capture) measures the distance to millions of surface points by firing a laser and timing or analysing its return. Sweeping the beam across a scene builds a dense point cloud — a 3D record where every point has coordinates and often colour and intensity. Where a total station measures points you choose, a scanner measures everything in view, capturing reality as-is.

This is the backbone of reality capture: documenting existing conditions for BIM, heritage, plant and infrastructure with a completeness no point-by-point method can match. A single scan can contain tens of millions of points; a project, billions.

Two ranging principles dominate, and the choice shapes range, speed and precision.

Time-of-flight scanners reach hundreds of metres and suit large outdoor sites. Phase-shift scanners are extremely fast and precise at short range, ideal for building interiors and industrial plant. Many modern instruments blend both. The same LiDAR principle drives airborne mapping covered in the Drone Surveying guide.

Terrestrial scanners give the best accuracy but you move and re-scan from many positions. SLAM (Simultaneous Localisation and Mapping) systems trade some accuracy for the speed of simply walking or driving through a site. Compare instruments in the surveying instruments database (laser scanner category) and makers like FARO, Leica and others in the manufacturers directory.

Each scan is captured in the scanner's own local coordinate frame. Registration is the process of aligning all the individual scans into one consistent point cloud. The main approaches:

• Target-based — place spheres or checkerboard targets visible from adjacent scan positions; the software matches them. Reliable and accurate, but slower to set up.
• Cloud-to-cloud — the software aligns overlapping geometry directly, no targets needed. Fast, and excellent when the scene has plenty of distinct shape.
• SLAM — mobile systems register continuously as they move, building the combined cloud in real time.

Registration quality is reported as a cloud-to-cloud error; keeping it within a few millimetres is the goal for precise work. Ensure scans overlap enough — typically 30–40% shared scene — for a robust alignment.

A registered cloud is internally consistent but still floating in space. Georeferencing places it in a real coordinate system by tying scan targets to control points measured with a total station or GNSS. This is what lets a scan combine with other survey data — see the Coordinate Systems guide for getting the datum right.

Two numbers describe a scan's fitness for purpose:

• Accuracy — how close each point is to truth (millimetres for terrestrial, centimetres for mobile).
• Point density / resolution — points per area at a given distance. Higher density resolves finer detail but multiplies file size.

Set resolution to the deliverable, not the maximum: scanning a façade for BIM needs far less density than capturing a casting for inspection. Over-scanning wastes time and storage.

The raw cloud is rarely the end product. Common deliverables built from it:

• Scan-to-BIM — modelling walls, structure and MEP from the cloud into an as-built BIM model.
• 2D drawings — plans, elevations and sections sliced directly from the cloud.
• Mesh / 3D model — a textured surface for visualisation, heritage or inspection.
• Deformation analysis — comparing scans over time, or a scan against design, to find movement or defects.

Scanning total stations bridge classic surveying and reality capture — see the Total Stations guide. Terminology is defined in the surveying glossary.