Point clouds are generated with scanning and sensing technologies and methods like LiDAR, structured-light scanning, time-of-flight cameras, sonar, radar and photogrammetry. These methods capture data about the geometry and spatial information for objects and environments by measuring distances, light patterns or wave reflections. The captured data is then processed and converted into a set of 3D points, each based on a specific position in space. Post-processing steps are often taken to refine the data, improve its accuracy, and translate it across trades and applications.

Point clouds can be used for a variety of purposes, primarily in industries like architecture, engineering, construction, entertainment, forensics, geospatial mapping, manufacturing and robotics. They offer a detailed 3D representation of real-world objects or environments, enabling complex tasks like as-built modelling, reverse engineering, virtual-reality experiences (US site), precise measurements and data visualisation. The use of point clouds makes many nuanced processes much more efficient, making them a versatile tool for capturing, analysing and using 3D spatial data.

An example of point cloud data is the representation of an object or environment captured using laser scanning technology or photogrammetry. Consider a building: To create a 3D point cloud of a building, a 3D scanner could be used to capture multiple data points and perspectives. Each point would represent a specific location in the building’s structure. The data points collected are then combined to form a point cloud that represents the 3D geometry of the building. In essence, point cloud data is any information gathered via laser scan or photogrammetry, then used to create the point cloud.

The difference between a point cloud and photogrammetry is in the way the data is captured and represented. A point cloud is a set of 3D points that represent the geometry of an object or space. Point clouds are typically generated using scanning technologies like LiDAR, or image-based methods like photogrammetry. Photogrammetry involves capturing multiple images of an object or environment from a variety of angles, then using software to calculate dimensions and plot precise points in 3D space.

A key difference between point clouds and mesh models is found in their digital representation and data structure. Point clouds are collections of individual points depicted in 3D space. Each point represents a specific location and can contain additional information like colour or intensity. In contrast, mesh models are surfaced-based representations built from interconnected polygons like triangles or quadrilaterals. They provide a closed surface that approximates the shape of the object, with edges, faces and vertices defining its structure. Point clouds retain the raw data and details, while mesh models offer a more compact and structured representation.

Point-cloud accuracy can vary based on the technology used for data capture, the scanning setup and the data processing methods applied. Advanced scanning technologies like LiDAR can achieve sub-centimetre or even sub-millimetre accuracy, while other methods may be slightly less precise. Factors like sensor limitations or movement within the scanned space can impact point-cloud accuracy as well. Accurate point clouds are often produced from multiple scans, and a thorough quality control process can minimise errors and ensure data is gathered effectively.