From Point Cloud to BIM: The Complete Workflow Explained

Stage 1: Site Capture

The process begins on site. A skilled operator places the scanner at multiple positions throughout the space — typically every 5–10 metres, depending on room geometry and required accuracy. Each scan position captures a complete 360° sphere of data.

For a standard office floor of 1,000 m², a thorough scan might involve 20–40 individual scan positions, each taking 2–5 minutes. The total site time is typically 2–4 hours, compared to a full day or more for traditional measured survey.

Stage 2: Registration

Back in the office, the individual scan positions must be registered — mathematically aligned into a single unified coordinate system. Modern software (Leica Cyclone, Autodesk ReCap, NavVis IVION) automates much of this process using overlapping geometry between adjacent scans.

The result is a single, georeferenced point cloud file that can be hundreds of gigabytes in size for a large building.

Stage 3: Point Cloud Processing

Before modelling begins, the raw point cloud is cleaned and optimised:

• Noise removal — eliminating stray points caused by reflective surfaces, moving objects, or scanner artefacts
• Decimation — reducing point density in areas where high resolution is not needed
• Classification — labelling points by surface type (floor, wall, ceiling, structural element)
• Segmentation — isolating specific building elements for individual modelling

Stage 4: 2D Drawing Production

For clients requiring 2D deliverables, the processed point cloud is sliced at standard heights (typically 1.0–1.2m for floor plans, vertical sections for elevations) to produce accurate linework. This is exported to AutoCAD or similar for annotation and finishing.

Accuracy at this stage is typically ±5mm for internal dimensions — sufficient for all planning, design, and construction purposes.

Stage 5: 3D BIM Modelling

For full BIM deliverables, a modeller works directly within the point cloud environment in Autodesk Revit or similar software. Walls, floors, ceilings, columns, beams, windows, doors, and M&E elements are modelled as intelligent BIM objects, using the point cloud as a precise reference.

The output is a fully coordinated BIM model that can be:

• Shared with design teams for refurbishment planning
• Used for clash detection in renovation projects
• Archived as a permanent as-built record
• Integrated with facility management systems

Stage 6: Quality Assurance

Before delivery, every model is checked against the original point cloud to verify that no significant deviations have been introduced during modelling. Dimensional spot-checks are documented and included with the deliverable package.

Typical Timescales

The speed advantage over traditional survey is most pronounced on medium and large projects, where conventional methods might take weeks to produce equivalent drawings.