TL;DR
A point cloud is a digital 3D model made up of millions—or billions—of individual data points, each with an X, Y, and Z coordinate. Think of it as a 3D photograph built from measurements rather than pixels. Point clouds are created by laser scanners, LiDAR sensors, or photogrammetry software. They are used in mining, construction, and manufacturing to create accurate digital replicas of physical objects and environments. A single laser scanner can capture 1-2 million points per second, creating a dataset dense enough to measure distances, check alignments, and detect clashes to millimetre precision.
Key Takeaways
- A point cloud is a set of millions of 3D coordinate points that together represent the shape and surface of a physical object or environment
- Point clouds are created by three main technologies: terrestrial laser scanning, LiDAR, and photogrammetry
- The most common file formats are E57, LAS, LAZ, PLY, and RCP/RCS—each suited to different software and workflows
- Point cloud processing software includes Autodesk ReCap, Leica Cyclone, Trimble RealWorks, CloudCompare, and Bentley ContextCapture
- A point cloud by itself is just data; its value comes from the analysis, modelling, and decision-making it enables
Table of Contents
- What is a point cloud? (explainer for non-technical readers)
- The simplest explanation: a 3D photograph made of dots
- How point clouds are created: three methods
- What does a point cloud look like?
- Point cloud file formats explained
- Point cloud software: processing and viewing
- What can you do with a point cloud?
- Point cloud accuracy: how precise is the data?
- Point cloud size and storage
- Point cloud vs 3D model: what's the difference?
- Frequently asked questions
- What to do next
The simplest explanation: a 3D photograph made of dots
Definition: A point cloud is a collection of millions of individual points in three-dimensional space, each defined by X, Y, and Z coordinates. Together, these points form a digital representation of a physical object's surface or an environment's geometry.
Imagine taking a photograph, but instead of recording colour for each pixel, you record a precise 3D position. Now imagine doing that millions of times, from multiple angles, until you have a complete three-dimensional picture of a room, a building, a mine site, or an entire city.
That is a point cloud.
Each point in the cloud has:
- X coordinate: Position east-west
- Y coordinate: Position north-south
- Z coordinate: Height or elevation
- Colour (optional): RGB values captured by an integrated camera
- Intensity (optional): The strength of the laser return signal, which can indicate surface material
| Property | 2D Photograph | Point Cloud |
|---|---|---|
| Basic unit | Pixel (X, Y + colour) | Point (X, Y, Z + optional colour) |
| Can you measure distance? | No (only relative) | Yes (millimetre precision) |
| Viewpoint | Single camera position | Any position in 3D space |
| Depth information | Approximate (via shadows) | Exact for every point |
| File size (typical) | 2-10 MB (JPEG) | 500 MB - 50 GB |
| Software to view | Any image viewer | Specialised 3D software |
How point clouds are created: three methods
Method 1: Terrestrial laser scanning (TLS)
A terrestrial laser scanner sits on a tripod and emits laser pulses in a 360-degree pattern. For each pulse, it measures the time taken for the light to return (time-of-flight) or the phase shift of the returning wave (phase-based). This gives the distance to the surface. Combined with the known angle of the laser emitter, the scanner calculates the 3D position of the point where the laser hit.
Typical specifications:
- Speed: 500,000 to 2,000,000 points per second
- Range: 0.5 m to 1,000 m (depending on scanner model)
- Accuracy: 1-3 mm at 10 m distance
- Multiple setups are registered together to cover large areas
Method 2: LiDAR (Light Detection and Ranging)
LiDAR is the airborne and mobile cousin of terrestrial laser scanning. A LiDAR sensor is mounted on a drone, aircraft, vehicle, or backpack and emits laser pulses downward or outward as the platform moves.
Types of LiDAR:
| Type | Platform | Typical Use | Point Density |
|---|---|---|---|
| Airborne LiDAR (ALS) | Fixed-wing aircraft or helicopter | Regional topography, forestry | 1-20 pts/m |
| Drone LiDAR | UAV / drone | Mine sites, quarries, construction | 50-500 pts/m |
| Mobile LiDAR (MLS) | Vehicle-mounted | Road corridors, rail, utilities | 100-1,000 pts/m |
| Static / Terrestrial | Tripod | Plant interiors, buildings, detail | 1,000-10,000 pts/m |
Method 3: Photogrammetry
Photogrammetry creates point clouds from photographs. A camera captures overlapping images from multiple positions, and software uses a technique called "structure from motion" to identify matching points across photos and triangulate their 3D positions.
How it works:
- Capture 50-500 overlapping photographs of the subject
- Software identifies common features across multiple photos
- Using camera positions and angles, the software calculates a 3D coordinate for each matched feature
- A dense point cloud is generated through a process called "dense matching"
| Method | Best For | Accuracy | Cost |
|---|---|---|---|
| Terrestrial laser scanning | Indoor plant, buildings, complex geometry | 1-3 mm | AUD 2,500-4,500/day |
| Drone LiDAR | Large sites, vegetation, terrain | 2-5 cm | AUD 2,000-3,500/day |
| Photogrammetry | Visual detail, inspection, small objects | 1-10 mm (close range) | AUD 1,500-3,000/day |
What does a point cloud look like?
When viewed in 3D software, a point cloud looks like a ghostly, ultra-detailed version of the scanned object. You can rotate it, zoom in, fly through it, and measure any distance between any two points.
At low zoom levels, a point cloud looks like a solid 3D model. At high zoom, you see the individual points—millions of tiny dots floating in space, each at its precisely measured location.
Most point cloud viewers allow you to:
- Rotate and orbit around the data
- Clip sections to look inside (like a CT scan)
- Measure distances, angles, and areas
- Compare the point cloud to a design model
- Extract specific features (pipes, walls, steel sections)
Point cloud file formats explained
Not all point cloud files are the same. Each format has strengths, weaknesses, and software compatibility considerations.
| Format | Extension | Best For | Compression | Software Support |
|---|---|---|---|---|
| E57 | .e57 | Universal exchange, archiving | Moderate | All major packages |
| LAS | .las | Standard for LiDAR data | None | GIS, CAD, specialist |
| LAZ | .laz | Compressed LiDAR exchange | High (7-20x) | Same as LAS |
| PTS/PTX | .pts / .ptx | Leica scanner native | None | Leica Cyclone |
| PLY | .ply | Research, 3D printing, simple exchange | Optional | Widely supported |
| RCP/RCS | .rcp / .rcs | Autodesk workflow (ReCap, Revit) | Good | Autodesk suite |
| XYZ/CSV | .xyz / .csv | Simple coordinate lists | None | Universal |
Key point: E57 has become the de facto standard for point cloud exchange between different software platforms. If you are sharing point cloud data with multiple stakeholders, request or deliver in E57 format. For Autodesk-centric workflows, RCP/RCS is more efficient.
Point cloud software: processing and viewing
| Software | Type | Best For | Price Range |
|---|---|---|---|
| Autodesk ReCap | Processing + viewing | Construction, BIM, Revit integration | AUD 400-600/year |
| Leica Cyclone | Processing + viewing | Industrial, mining, complex registration | AUD 8,000-15,000/year |
| Trimble RealWorks | Processing + viewing | Construction, Trimble ecosystem | AUD 5,000-10,000/year |
| Bentley ContextCapture | Processing | Large-scale reality modelling | AUD 10,000+/year |
| CloudCompare | Viewing + basic processing | Free alternative, format conversion | Free (open source) |
| Faro Scene | Processing + viewing | Faro scanner users, forensic | AUD 4,000-8,000/year |
| 12d Model | Processing + design | Civil engineering, earthworks | AUD 5,000-10,000/year |
| Global Mapper | Viewing + analysis | GIS integration, terrain analysis | AUD 800-1,500/year |
What can you do with a point cloud?
Point clouds are not just visualisations. They are measurement databases that enable specific engineering and operational tasks.
| Application | What You Do | Industry |
|---|---|---|
| As-built documentation | Compare point cloud to design model; identify deviations | Construction, mining |
| Clash detection | Overlay design model on existing conditions to find interferences | Engineering, construction |
| Dimensional control | Measure distances, clearances, and alignments to specification | Manufacturing, mining |
| Volume calculations | Calculate stockpile volumes, cut-and-fill quantities | Mining, quarrying |
| Deformation monitoring | Compare point clouds captured at different times to detect movement | Structural engineering |
| BIM creation | Convert point cloud into a structured Building Information Model | Architecture, facilities management |
| Reverse engineering | Create CAD models of existing objects from scan data | Manufacturing, heritage |
| Quality inspection | Compare manufactured parts to design specifications | Manufacturing |
| Heritage recording | Create permanent digital records of cultural heritage sites | Conservation, government |
Point cloud accuracy: how precise is the data?
The accuracy of a point cloud depends on the capture method, the equipment, the environment, and the processing.
| Capture Method | Typical Accuracy | Factors Affecting Accuracy |
|---|---|---|
| Terrestrial laser scanning | 1-3 mm at 10 m | Scanner calibration, range, surface reflectivity, registration error |
| Drone LiDAR | 2-5 cm | GNSS/IMU quality, flight altitude, ground control, vegetation |
| Airborne LiDAR | 10-30 cm | Altitude, atmospheric conditions, GNSS quality |
| Photogrammetry | 1-10 mm (close range) | Camera quality, overlap, ground control, lighting |
| Mobile scanning (SLAM) | 1-5 cm | Traverse length, loop closure, feature-richness of environment |
Key factors that reduce accuracy:
- Range: Accuracy degrades with distance from the scanner
- Surface material: Dark, shiny, or transparent surfaces reflect laser poorly
- Registration error: Misalignment between multiple scan positions
- Movement: Vibration or movement during scanning
- Environmental conditions: Dust, rain, and extreme temperature affect sensors
Point cloud size and storage
Point clouds are large. Understanding storage requirements is important for project planning.
| Project Type | Typical Point Count | Raw Data Size | Processed Size |
|---|---|---|---|
| Single room (office) | 10-50 million | 2-5 GB | 500 MB - 2 GB |
| Building floor (500 m) | 100-500 million | 10-50 GB | 5-20 GB |
| Processing plant section | 1-5 billion | 100-500 GB | 50-200 GB |
| Full mine site (drone LiDAR) | 500 million - 2 billion | 50-200 GB | 20-100 GB |
| Urban block (terrestrial) | 5-20 billion | 500 GB - 2 TB | 200 GB - 1 TB |
Compression techniques like octree encoding and LAZ compression can reduce file sizes by 50-90% with no loss of precision.
Point cloud vs 3D model: what's the difference?
This is a common source of confusion. A point cloud and a 3D model are related but different.
| Property | Point Cloud | 3D Model (CAD/BIM) |
|---|---|---|
| Representation | Millions of discrete points | Continuous surfaces, lines, and solids |
| Intelligence | None—just coordinates and colour | Object classes, relationships, metadata |
| Editability | Can be filtered and clipped; not easily modified | Fully editable parametric geometry |
| File size | Very large (GB to TB) | Smaller (MB to GB) |
| Creation | Captured from reality | Designed in software or extracted from point cloud |
| Software | ReCap, Cyclone, CloudCompare | Revit, AutoCAD, 12d, Tekla |
The typical workflow is: capture point cloud → process and clean → extract 3D model → use model for design and analysis.
The point cloud is the raw measurement. The 3D model is the engineering interpretation.
Frequently asked questions
What is the difference between a point cloud and a mesh?
A point cloud is a set of discrete points with no connectivity between them. A mesh connects those points into triangles (or polygons) to create continuous surfaces. A mesh is generated from a point cloud through a process called "surface reconstruction." Meshes are smaller, smoother, and more suitable for visualisation and 3D printing. Point clouds retain the raw measurement precision.
How much does it cost to get a point cloud of my facility?
Costs depend on facility size and complexity. A single processing plant section (50 m x 50 m) typically costs AUD 3,000-8,000 to scan and deliver as a point cloud. A full building (2,000 m) ranges from AUD 8,000-20,000. A full industrial site with drone and ground capture can range from AUD 25,000-80,000. ISS provides fixed-fee quotes for point cloud capture projects.
Can I use a point cloud in Revit or AutoCAD?
Yes. Autodesk ReCap converts point clouds into RCP/RCS files that load directly into Revit and AutoCAD. You can snap to points, create sections, and trace over the cloud to build BIM elements. Most ISS deliverables include ReCap-compatible files.
How long does it take to capture a point cloud?
Field time varies: a single room takes 30-60 minutes; a building floor takes 2-4 hours; a processing plant section takes 4-8 hours; a full site takes 1-5 days. Processing time is typically equal to or double the field time.
Is a point cloud admissible as a legal record?
Point clouds captured by a registered surveyor using calibrated equipment, with documented methodology and quality control, can form part of a legal survey record. The surveyor's certification and methodology documentation are what make it legally defensible, not the point cloud format itself.
What to do next
Point clouds have become the standard way to capture and document physical reality for industrial projects. If you are managing assets, planning construction, or retrofitting existing facilities, a point cloud gives you a measurable, permanent digital record.
- Define your purpose. Do you need a point cloud for clash detection, as-built documentation, volume calculations, or a digital twin? The purpose determines the capture method, accuracy, and deliverable format.
- Estimate your scope. How large is the area? How complex is the geometry? How dense does the point cloud need to be? These factors drive equipment selection and cost.
- Specify your deliverable format. Will you process the point cloud yourself, or do you need the surveyor to provide a finished model? Specify E57, RCP, LAS, or another format upfront.
Call ISS on 0407 057 015 to discuss your point cloud requirements. We will recommend the right capture method—laser scanning, drone LiDAR, or photogrammetry—and deliver a point cloud matched to your accuracy needs and software workflow.