3D rendering glossary
Point cloud: what it is, how it is captured and what it is used for
A point cloud is a set of millions of points with X, Y and Z coordinates that describes the surface of an object, a building or a terrain. It is captured with a laser scanner, LiDAR or photogrammetry, and each point can also store colour and intensity. It is the basis for creating accurate 3D models of what already exists.
How a point cloud is generated
There are three main capture methods, each suited to a particular situation:
| Method | How it works | When to use it |
|---|---|---|
| Terrestrial laser scanner (TLS) | A tripod-mounted unit emits laser pulses and measures the distance to each surface | Facades, interiors, heritage, industrial buildings; maximum precision |
| Aerial or drone LiDAR | A laser sensor mounted on an aircraft or drone sweeps the terrain from the air | Topographical surveys, large plots, areas with vegetation |
| Mobile scanner (SLAM) | The device is carried by hand, in a backpack or on a vehicle and positions itself as it moves | Long routes, complex buildings, rapid surveys |
| Photogrammetry | Software cross-references hundreds of overlapping photographs and calculates the position of each point | Facades, roofs, objects, lower-cost drone surveys |
Several captures (stations) are combined into a single coordinate system through a process called registration, supported by targets, control points or geometric overlaps. The result is saved in formats such as E57, LAS/LAZ, PLY, PTS or the native formats of each manufacturer.
In practice: Render Fabrik is a 3D architectural rendering studio working with clients in Brazil and Europe, with a quote within 24 hours.
From point cloud to 3D model
A point cloud is not yet a model: it is a dense sample of the surface. To use it in design or visualisation, the usual workflow is:
- Cleaning: remove noise, people, cars and reflections captured by mistake.
- Classification: separate ground, vegetation, buildings and other elements.
- Modelling: draw walls, slabs, joinery and roofs over the cloud (scan to BIM), or generate a polygon mesh directly from the points.
- Verification: compare the model with the cloud to measure deviations.
Applications in architecture, construction and industry
- Refurbishment and renovation: exact survey of an existing building before designing, without relying on old drawings that often do not match reality.
- Heritage: documentation of churches, castles and historic facades with full detail.
- Construction control: compare what is built with the BIM model to detect deviations.
- Topography: digital terrain models for developments, roads and volume studies.
- Industry: plants, pipework and installations where new equipment must fit into already occupied spaces.
- Visualisation: integrate a new project into its real environment, with the exact volume of neighbouring buildings.
Advantages and limitations
Advantages: very high precision, rapid capture of complex geometries, complete information that allows measurement later without returning to site, and fewer errors than manual measurement with tape or laser distance meter.
Limitations:
- Very large files that require powerful computers and specialised software.
- Shadow areas where the scanner cannot reach, which require adding stations or photographs.
- Problematic surfaces: glass, mirrors, water and very dark materials return poor or erroneous data.
- The cloud has no "intelligence": a wall is a set of points until someone models it as a wall.
What to ask for in a point cloud survey
- Accuracy and density: whatever is needed for the intended use; a volume study does not require the same as a refurbishment project.
- Delivery format: E57 or LAS for exchange, and the cloud already registered and clean.
- Coordinate system: georeferenced if it is to be combined with mapping or topography.
- Photographs or panoramic images: they help identify materials and elements when modelling.
Common software
For registration and processing: Leica Cyclone, Trimble RealWorks, FARO Scene or Autodesk ReCap. For project work: Revit, Archicad or AutoCAD with linked point clouds. For analysis and cleaning, CloudCompare, free and open source, is widely used. In photogrammetry, Agisoft Metashape, RealityCapture and Pix4D stand out.
The point cloud in a visualisation studio
In architectural visualisation, a point cloud or the model derived from it allows you to place a new building in its real context with accuracy, or to reproduce an existing interior to show a refurbishment. If you work in Spain and have a scanned survey, you can visit our page for Spain, see examples in the architectural visualisation section or write to us via contact.
Frequently asked questions
What is the difference between a point cloud and a 3D mesh?
A point cloud is a set of separate points with no connection between them. A mesh joins those points into faces (usually triangles) and forms a continuous surface that can be textured and rendered.
How accurate is a point cloud?
It depends on the equipment and method. A terrestrial laser scanner offers accuracies of a few millimetres, while drone photogrammetry is usually in the order of centimetres, depending on flight height and photo quality.
What is scan to BIM?
It is the process of converting a point cloud into a BIM model with real building elements: walls, slabs, columns, windows and installations, each with its own information.
Can a point cloud be rendered directly?
Some programs display and render it as points, which is useful for technical images or visual effects. For a photorealistic image, the geometry must be modelled or meshed and materials assigned.