BIM IQ

Laser Scanning in Construction: Methods, Accuracy and When It Is Worth It

By Kush Agarwal, President and CEO, YelloSKYE. Published 2026-10-20.

Executive Summary

• The Question: What is laser scanning, how do the different kinds of scanner compare, and when is a scan worth the cost on a construction project?
• The Answer: A laser scanner measures surfaces directly with laser light and produces a dense point cloud. Tripod, mobile and aerial systems trade precision against speed and coverage. Scanning earns its cost when a decision depends on exact geometry; for visual records and open terrain, 360 capture and drone photogrammetry are usually the better fit.
• The YelloSKYE Angle: BIM IQ turns point clouds from the major scanner brands, and from drone surveys, into as-built BIM models and checks them against design, with every capture held on one timeline in the DEX platform.

Laser scanning in construction is the use of a laser instrument to measure the surfaces of a building, structure or site directly, producing a point cloud: a dense set of 3D coordinates that records the geometry exactly as it stands. Teams use it when they need to know where something really is, not just what it looks like, which makes it the usual starting point for as-built models, checks against design and renovation work on existing buildings.

At YelloSKYE, scan data feeds BIM IQ, our scan-to-BIM solution. This guide explains how scanners measure, how terrestrial, mobile and aerial systems differ, how scanning compares with photogrammetry and 360 capture, what controls the accuracy of a scan and how to judge whether a scan is worth commissioning. For how scan data and the design model are kept in step afterwards, see our guide to reality capture and BIM.

How a laser scanner measures

A scanner sends out laser light, catches the reflection and works out how far away the surface is. It does this while rotating or sweeping, so every reading comes with a direction as well as a distance. Combine the two and each reading becomes a point with X, Y and Z coordinates.

Two measuring methods are common:

• Time of flight. The scanner times how long a pulse takes to return. This works over long distances and suits large halls, facades and outdoor structures.
• Phase shift. The scanner sends a continuous beam and compares the phase of the outgoing and returning light. It is very fast and very dense at shorter ranges, which suits interiors and plant rooms.

Most scanners also carry a camera, so each point can be given a colour. That makes the cloud readable to people who are not survey specialists, although the colour is decoration: the measurement comes from the laser.

One scan position can only record what it can see. A column hides the wall behind it, and the top of a beam is invisible from the floor. The operator therefore moves the scanner around the space and joins the positions afterwards in a step called registration. The quality of that join matters as much as the quality of the instrument.

Three ways to carry a scanner

The same principle is packaged in very different ways depending on whether the priority is precision, speed or coverage.

Terrestrial scanning

The scanner stands still on a tripod for each setup, then moves to the next position. Because it is stationary while measuring, it gives the cleanest and most precise data of the three. It is the standard choice for structural verification, congested service areas, heritage fabric and anything that will be modelled to a tight tolerance. The cost is time: every setup has to be planned, levelled and later registered.

Mobile scanning

Handheld units, backpacks and wheeled trolleys scan continuously while the operator walks. They work out their own movement using a technique called SLAM (simultaneous localisation and mapping), which compares each new slice of data with what the scanner has already seen. A whole floor can be covered in a fraction of the time a tripod survey needs. The trade-off is drift: small errors in the scanner's estimate of its own path build up over a long walk, so mobile data is usually less precise and benefits from control targets and closed loops in the route.

Aerial LiDAR

A LiDAR sensor mounted on a drone or aircraft scans the ground from above while satellite positioning and a motion sensor track where the platform is. A single pulse can return several echoes as it passes through vegetation, so aerial LiDAR can model bare ground under tree cover in a way photographs cannot. It suits corridors, wooded sites and large terrain surveys. It is not a tool for building interiors or element-level structural checks.

Laser scanning vs photogrammetry vs 360 capture

Laser scanning is often discussed alongside two other capture methods. They overlap, but each answers a different kind of question.

Photogrammetry reconstructs geometry from many overlapping photographs. It needs good light and surfaces with visible texture, and its geometry is calculated rather than measured. From a drone with ground control, it is the most efficient way to map open sites, earthworks and roofs, and it also produces an orthomosaic, which a scanner does not. Indoors, on plain painted walls or in dim plant rooms, it struggles.

360 capture records spherical images along a walking route and pins them to a floor plan. It is quick, cheap enough to repeat every week and easy for site staff to operate. It shows what was there on a date, which makes it ideal for progress records, snagging and evidence before ceilings close. It is not a measuring tool: you can see that a duct was installed, but not confirm its height to the millimetre. Our guide to 360 cameras for construction covers how to run that routine.

The practical rule is to use the cheapest method that answers the question with the certainty the decision requires. Most projects end up with all three, each tied to the same survey control so the datasets line up. Our guide to hybrid construction surveys explains how that control is set up.

What decides the accuracy of a scan

The number on a scanner's specification sheet describes the instrument under test conditions. The accuracy of the dataset you receive depends on several other things, and they are worth understanding before you set a tolerance in a scope.

Range and angle

Precision falls as the distance to the surface grows, and readings taken at a glancing angle are noisier than readings taken square on. A well-planned scan places setups so that critical elements are measured at moderate range from more than one direction.

Surface material

Matt, light-coloured surfaces return a clean signal. Glass lets the beam through, polished metal and water can reflect it away, and very dark surfaces return little light. These areas show up as gaps or false points and may need extra setups or manual checks.

Registration

Joining scan positions together, either with physical targets or by matching overlapping geometry, introduces its own error. A registration report lists how well each scan fits its neighbours. If a provider cannot produce one, you cannot tell how much to trust the cloud.

Control and coordinates

A cloud can be internally consistent and still sit in the wrong place. Tying scans to the project's survey control places them in the same coordinate system as the design model and other datasets, which is what makes a deviation report meaningful.

Movement and conditions

Vibration from nearby plant, people walking through the scan, dust and heat shimmer all add noise. For mobile systems, the length and shape of the route matter because drift grows between control points.

The right question to ask is not "how accurate is the scanner?" but "what tolerance does each element need, and how will the delivery prove it was met?"

When laser scanning is worth it

Scanning costs more per square metre than photographs, so it should be targeted at the moments where an exact measurement changes a decision.

Two signs that a scan is not the right purchase: nobody can name the decision the data will support, or nobody on the team has the software or skills to open a point cloud. In the second case, ask for the deliverable you will actually use, such as a model, drawings or a deviation report, rather than the raw cloud.

From point cloud to a usable deliverable

A raw point cloud is a measurement, not an answer. Turning it into something a design or site team can use follows a consistent sequence.

1. Register and check. Individual scans are joined and tied to control, with a report on the fit.
2. Clean. Temporary objects, people, reflections and noise are removed so they are not mistaken for building elements.
3. Model or compare. The cloud is either modelled into BIM elements at an agreed level of development, or laid against the design model to measure where built elements differ from it.
4. Deliver in the team's formats. Clouds, models and reports go to the people who need them in files their software reads.

BIM IQ accepts point clouds from Leica, FARO, Trimble and Matterport scanners, as well as drone photogrammetry outputs, in formats including .e57, .las, .rcp, .pts, .xyz and .pcg. It delivers as-built models from LOD 200 to LOD 400, exports to DWG, DXF, RCP, LAS, OBJ and FBX for tools such as Revit and Navisworks, and overlays the scan-derived model on the design model to flag misaligned elements. On the DEX platform, laser scans sit alongside drone, 360 and fixed-camera captures, so a scan of a slab can be viewed with the weekly walks that came before and after it.

Two of our use cases show the main patterns. In scan to BIM for renovation, an existing building is scanned and modelled so architects and consultants start from measured conditions rather than legacy drawings. In as-built vs BIM clash detection, the structure is scanned after a slab pour or steel erection and compared with the design model before MEP contractors arrive. Where the question is visual rather than dimensional, our interior fit-out work shows the 360 alternative.

Planning a scan: a checklist

• Name the decision. Write down what the scan must prove and who will act on it.
• Choose the system to match. Terrestrial for tight tolerances, mobile for speed over large interiors, aerial LiDAR for terrain and vegetated ground.
• Set tolerances by element. Structure, services and finishes rarely need the same accuracy.
• Fix the coordinate system. Give the provider the project control and require scans to be tied to it.
• Time it to the milestone. Scan after the work is complete and before the next trade covers it.
• Flag difficult surfaces. Glass, polished metal and dark finishes need extra setups or checks.
• Specify the deliverables. Point cloud format, model LOD, deviation report and registration report, in the formats your team already uses.
• Keep it with the rest of the record. Store the scan where the site, design and owner teams can open it alongside other captures.

Explore BIM IQ