A point cloud is a dataset made up of millions (or billions) of individual 3D points, where each point has an X, Y, and Z coordinate representing a precise location in physical space. Together, these points form a detailed three-dimensional representation of a real-world object, structure, or landscape. Point clouds are the foundational output of LiDAR scanning, photogrammetry, and terrestrial laser scanning — and they are how surveyors capture the physical world digitally.
If you are a surveyor, you work with point clouds every day. If you are a client who has just received a point cloud from your surveyor, this guide will help you understand what you are looking at and why it matters.
What does a point cloud look like?
Imagine standing in a room and placing a tiny coloured dot on every visible surface — every wall, floor, ceiling, piece of furniture, doorframe. Now imagine placing those dots every 2 millimetres, across every surface, with each dot recording its exact 3D position. That collection of dots is a point cloud.
When viewed in software, a point cloud looks like a three-dimensional scene made of particles rather than solid surfaces. You can rotate around it, zoom into details, and fly through it. At a distance, it looks like a photograph. Up close, you can see the individual points.
The density of points determines the level of detail. A dense point cloud (thousands of points per square metre) resolves fine features — cracks, edges, small objects. A sparse point cloud shows the general shape but misses fine detail.
How are point clouds captured?
There are three primary methods for capturing point clouds, each suited to different scenarios.
LiDAR (Light Detection and Ranging)
LiDAR scanners emit laser pulses and measure the time it takes for each pulse to bounce back from a surface. This time-of-flight measurement gives a precise distance, and by sweeping the laser across a scene, the scanner builds up millions of distance measurements — each becoming a 3D point.
Airborne LiDAR is mounted on aircraft or drones and scans the ground from above. It is excellent for terrain mapping, forestry, corridor surveys (powerlines, roads), and large-area topographic surveys. Airborne LiDAR can penetrate vegetation canopy, making it the preferred technology for mapping ground surfaces under tree cover.
Terrestrial LiDAR uses tripod-mounted scanners (from manufacturers like FARO, Leica, and Trimble) that scan a scene from fixed positions. These scanners capture extremely dense point clouds — often hundreds of millions of points per scan position — and are used for building surveys, heritage documentation, industrial plants, and forensic scene capture. For more on terrestrial scanning workflows, see our guide on sharing FARO and Leica scan data.
Mobile LiDAR is mounted on vehicles, backpacks, or handheld devices and captures point clouds while moving. Used for road surveys, indoor mapping, and rapid site documentation.
Photogrammetry
Photogrammetry derives 3D point clouds from overlapping photographs. By identifying the same feature in multiple photos taken from different angles, software calculates the 3D position of that feature through triangulation.
In drone survey workflows, photogrammetry is the most common method: the drone flies a grid pattern, capturing hundreds of overlapping photos, and software like Pix4D or Agisoft Metashape processes them into a dense point cloud.
Photogrammetry point clouds have RGB colour by default (because the source data is photographs), which makes them visually rich. However, they cannot penetrate vegetation and are less effective in low-texture environments (blank walls, snow, water).
Structured light scanning
Structured light scanners project patterns of light onto surfaces and use cameras to measure how the patterns deform. This provides very high-resolution point clouds at short range and is used primarily for small objects, industrial inspection, and heritage artefact documentation.
Properties of a point cloud
Every point in a point cloud has a position (X, Y, Z). Most point clouds also carry additional attributes per point:
Coordinates (X, Y, Z)
The fundamental property. Each point has three coordinate values that define its position in 3D space. For georeferenced point clouds (most survey data), these coordinates are in a known coordinate reference system (e.g., MGA2020, OSGB36, NAD83) and represent real-world positions.
RGB colour
If the point cloud was captured by photogrammetry or by a LiDAR scanner with an integrated camera, each point has Red, Green, and Blue colour values. This means the point cloud looks like a 3D colour photograph when rendered.
Intensity
LiDAR scanners record the strength of the returned laser pulse for each point. This intensity value depends on the surface material — asphalt, concrete, vegetation, metal, and water all reflect laser light differently. Intensity data is valuable for classification and material identification.
Classification
Points can be assigned a classification code that identifies what they represent. The ASPRS (American Society for Photogrammetry and Remote Sensing) standard defines classes including:
| Class code | Description |
|---|---|
| 2 | Ground |
| 3 | Low vegetation |
| 4 | Medium vegetation |
| 5 | High vegetation |
| 6 | Building |
| 9 | Water |
| 17 | Bridge deck |
| 64-255 | User defined |
Classification is typically performed automatically during processing and refined manually. It is essential for extracting ground surfaces (DTM generation), building footprints, and vegetation analysis.
Return number
Airborne LiDAR pulses can generate multiple returns — the laser hits a tree canopy (first return), passes through gaps, and hits the ground (last return). Return number data allows separation of canopy from ground surface, which is why LiDAR can “see through” vegetation.
Point density
Point density (points per square metre) determines the level of detail:
| Density | Typical use | Detail level |
|---|---|---|
| 1-5 pts/m2 | Large-area airborne LiDAR | Terrain, broad features |
| 10-50 pts/m2 | Drone photogrammetry, UAV LiDAR | Site-level detail, structures |
| 100-1,000 pts/m2 | Terrestrial scanning | Millimetre-level detail |
| 1,000+ pts/m2 | Close-range terrestrial, structured light | Sub-millimetre, inspection |
What are point clouds used for?
Surveying and engineering
The primary use case. Point clouds provide a complete 3D record of a site at a moment in time. From a point cloud, you can extract measurements, generate contours, create cross-sections, calculate volumes, and produce CAD drawings. They are used for:
- Topographic surveys
- Volumetric calculations (stockpiles, earthworks)
- As-built documentation
- Progress monitoring (comparing point clouds over time)
- Boundary and feature surveys
Construction
Construction teams use point clouds to compare as-built conditions against design models (BIM). A scan of a partially completed building, overlaid on the design model, reveals deviations — columns that are 20mm out of position, floors that are not level, services that deviate from the plan.
Heritage and preservation
Point clouds provide millimetre-accurate records of heritage structures — churches, castles, monuments, archaeological sites. These records serve as documentation for preservation and restoration, and as a baseline against which future deterioration can be measured.
Mining and resources
Stockpile volume measurement, pit progression monitoring, and geological mapping all rely on point cloud data. Regular drone surveys of a mine site produce a time series of point clouds that track extraction progress and volume changes.
Forestry
Airborne LiDAR point clouds penetrate the canopy and reach the ground, allowing simultaneous mapping of terrain and vegetation structure. Foresters use this data for tree height measurement, canopy density analysis, biomass estimation, and harvest planning.
File formats for point clouds
Point clouds are stored in several formats, each with different characteristics:
| Format | Use case | Compression | Notes |
|---|---|---|---|
| LAS | Universal exchange format | None | ASPRS standard, large files |
| LAZ | Compressed LAS | Lossless (5-10x) | Industry standard for delivery |
| E57 | Terrestrial scanning | Basic | ASTM standard, multi-scan |
| PLY | Research, 3D printing | Optional | Simple format, good mesh support |
| PCD | Point Cloud Library | Optional | Common in robotics/research |
| COPC | Cloud-optimised | LASzip | Streaming-ready LAZ variant |
| Potree | Browser viewing | Octree-based | Converted for web delivery |
For most survey workflows, LAZ is the standard deliverable format. It is well-supported, compact, and universally understood in the industry.
How to view a point cloud
If you are a surveyor (desktop software)
You already have the tools: CloudCompare (free and open source), QGIS, Global Mapper, or your processing software (Metashape, Pix4D, RealWorks, Cyclone). These tools handle large point clouds well and provide measurement, classification, and export capabilities.
If you are a client (no specialist software)
This is where it gets difficult. If your surveyor sends you a LAZ file, you need software to open it. CloudCompare is free, but it is not intuitive for non-technical users, and asking a project manager to install and learn a point cloud viewer is a significant friction point.
The better experience is browser-based point cloud viewing. Technologies like Potree can render massive point clouds in a standard web browser — Chrome, Firefox, Safari, Edge — with no software installation. You click a link, the point cloud loads, and you can rotate, zoom, and explore.
This is how Swyvl works: your surveyor uploads the point cloud, and you receive a branded link that opens it directly in your browser. You see the data immediately. You can zoom into areas of interest, rotate to any angle, and switch between colour modes (RGB, intensity, elevation, classification). If you need the raw file, you can download it too.
For clients: your surveyor just sent you a point cloud
If you have received a point cloud from your surveyor and are trying to understand what you are looking at, here is a quick orientation:
What you are seeing: A three-dimensional representation of the physical site, made up of millions of individual coloured dots. Every dot is a real measurement of a real surface.
How to navigate: Left-click and drag to rotate. Right-click and drag to pan. Scroll to zoom. These controls work in most point cloud viewers, including browser-based ones.
What to look for: Zoom into areas of interest. Rotate to see features from different angles. If your surveyor has classified the point cloud, you may be able to toggle between colour views — RGB (photo colour), elevation (height-based colour gradient), or classification (ground, vegetation, buildings shown in different colours).
What you can do with it: Depending on the viewer, you may be able to take measurements (point-to-point distance, area, elevation differences), create cross-sections, or annotate specific features. Ask your surveyor what measurement tools are available in the viewer they provided.
If you need the raw data: You can always download the LAZ file from the viewer interface. This is useful if you need to import the data into your own CAD or GIS software, or if you want to archive the raw deliverable.
Why point clouds matter
Point clouds have fundamentally changed surveying. A traditional survey captures discrete points — a few hundred or a few thousand carefully selected positions. A point cloud captures everything — every surface, every edge, every feature, with nothing left out. If you realise six months later that you need a measurement you did not take, it is probably in the point cloud.
This completeness is what makes point clouds valuable as a record. A comprehensive drone survey captured as a point cloud is not just the measurements you thought you needed — it is every measurement you might ever need from that site, at that moment in time.
The challenge has shifted from capturing the data to delivering it. Point clouds are large, specialised files that most end clients cannot open. Solving that delivery problem — making point clouds accessible to anyone with a browser — is what turns a specialist dataset into a universally useful spatial record.