A 3D point cloud is a set of points in space, each defined by its X, Y, and Z coordinates, that together describe the surface of an object, a building, or an entire site. It is the raw output of a 3D scan, before any modeling.
This page gives a short, clear definition. For how point clouds are created and used in detail, see the full guide linked below.
Definition of a 3D point cloud
Technically, a point cloud is a discrete set of data points in a coordinate system. Each point carries at least its position, and often several extra attributes captured by the scanner.
- X, Y, Z: the mandatory coordinates that place the point in 3D space.
- RGB color: added when the scanner has a camera, for a realistic look.
- Intensity: the strength of the returned laser pulse, revealing materials.
- Normals: the surface direction at the point, useful for meshing.
- Classification: a label (ground, vegetation, building) added in post-processing.
Not to be confused with
A 3D point cloud has nothing to do with the statistical scatter plot (a 2D chart in Excel or Python). Same words, unrelated meaning. This page is about the 3D scanning sense.
Point cloud vs mesh vs BIM model
A point cloud is not a 3D model. It is the raw material from which models are built. The three are easy to tell apart.
- Point cloud: measured reality, nothing invented.
- Mesh: a continuous surface, good to look at, not intelligent.
- BIM model: objects that know what they are, for design and management.
| Object | What it is | Typical use |
|---|---|---|
| Point cloud | A set of measured points | Measuring, inspecting, as-built reference |
| Mesh | Points connected into surfaces | Visualization, 3D printing, simulation |
| BIM model | Smart objects with data (walls, pipes) | Design, coordination, facility management |
Swipe the table horizontally to see all columns.
Types of point clouds
Dense vs sparse
A dense cloud has many points close together, for fine detail. A sparse cloud has fewer points, lighter to handle but less precise.
Colored or not
A colored cloud carries RGB values from a camera, for a photo-realistic look. A raw cloud may be shown by intensity or elevation instead.
Structured or unstructured
- Structured: points organized in a grid (typical of terrestrial scans), easier to process.
- Unstructured: points with no grid (typical of mobile or aerial scans), more flexible.
- Registered: several scans aligned into one coherent cloud.
View a point cloud in your browser
Upload an E57, LAS, LAZ, RCS, RCP, or LGSx file and explore it online, up to 1 TB per file.
How to read a point record
- Position: X, Y and Z locate the sample in a local or georeferenced coordinate system.
- Appearance: RGB values may come from an integrated or external camera.
- Return: intensity records the strength of the received laser signal.
- Interpretation: classification can label ground, vegetation, structures or another semantic class.
- Context: scan position, timestamp or return number can preserve how the measurement was acquired.
Density, precision and accuracy are not the same
These terms describe different properties of the data. A cloud can contain many points and still be poorly located. For creation methods, formats and professional uses, continue with the complete guide what is a point cloud.
- Density: how closely the samples are spaced on a surface.
- Precision: how consistently repeated measurements agree with one another.
- Accuracy: how close the measured position is to its true position.
- Resolution: the smallest detail the acquisition system can distinguish.
- Completeness: whether the required surfaces are covered without critical gaps.
A point cloud is the measurement, not the model. Everything you build afterward, mesh, plan, or BIM, starts from those points.
Share a point cloud with your clients
Send a link, your client sees the scan in 3D in the browser, with nothing to install.
A 3D point cloud is a set of measured points (X, Y, Z plus attributes) that samples a real surface. It is distinct from a mesh and a BIM model. Density describes point spacing, precision describes repeatability, and accuracy describes closeness to the true position.
Frequently asked questions
It is a set of points in space, each with X, Y, Z coordinates, that together map the surface of a real object or building. It is what a 3D scanner produces before any modeling.
No. A point cloud is raw measured points. A 3D model (mesh or BIM) is built from those points by connecting them into surfaces or turning them into smart objects.
Every point has X, Y and Z coordinates. It may also contain RGB color, laser intensity, a normal, a classification, a timestamp, scan position or return number.
No. Density describes how close the points are to one another. Accuracy describes how close their measured positions are to the true positions. A dense cloud can still contain systematic errors.
A dense cloud samples a surface with closely spaced points and can preserve finer detail. A sparse cloud uses fewer points, which reduces weight but also the amount of geometric detail available.
The complete What Is a Point Cloud guide covers LiDAR, photogrammetry, acquisition methods, file formats, professional uses, viewing and sharing. This page remains focused on the technical definition.
Related articles
Scan-to-BIM services or in-house software?
An external provider, desktop software or ATIS.cad in your browser: compare three ways to turn survey data into a usable BIM model.
How to Create a Point Cloud: Methods and Steps
How to create a point cloud step by step: laser scanning, photogrammetry, or mobile mapping, then registration, cleaning, export, and sharing.
Point Cloud File Formats: E57, LAS, LAZ, RCS, LGSx
The main point cloud file formats explained: E57, LAS, LAZ, RCS, RCP, LGSx. Open vs proprietary, which to use when, and how to convert.