Practical Geo Scan guidance on Point Cloud Vs CAD Vs BIM for architects, developers and project teams.
It is tempting to think of point cloud, CAD and BIM as three versions of the same thing. In practice, each represents the surveyed building differently.
A point cloud is close to the measured source. A CAD drawing selects and interprets that source into two-dimensional linework. A BIM model interprets it further into three-dimensional objects with defined relationships and, where scoped, associated information.
That progression can make CAD and BIM easier to use for particular design tasks, but interpretation also means decisions have been made about what the measured data represents.
This is why the deliverable should be selected around its intended use rather than simply ordering the most sophisticated-sounding output.
A point cloud is a collection of spatially positioned measured points representing visible surfaces captured during a survey. Terrestrial laser scanners can collect dense three-dimensional observations from many instrument positions, which are then registered into a coherent dataset.
Depending on the capture and export, points can also contain RGB colour, intensity and other attributes.
The principal strength of the point cloud is that it preserves a large amount of measured spatial information. A design team can inspect geometry that may not have been selected for a drawing, take additional measurements and revisit areas of the survey without asking for every possible feature to be drafted in advance.
Its limitation is equally important: a point cloud is not automatically a set of architectural objects. It does not inherently tell the user that a group of points is a wall, beam, window or pipe. That interpretation is performed by people or downstream software.
See Point Cloud Surveys and Point Cloud Accuracy & Registration for the underlying survey workflow.
Scan to CAD converts measured survey information into conventional two-dimensional drawings.
Typical outputs include:
The drafting process selects the geometry relevant to the agreed drawing scale and scope. A wall might be represented by clean linework even though the source point cloud contains millions of points across its visible surfaces.
That makes CAD efficient for architects and engineers who need a familiar drawing base. It also means the CAD file should not be mistaken for the complete measured dataset. Information outside the drawing scope may still exist in the point cloud without appearing in the DWG.
Scan to BIM uses measured survey data as the basis for a structured three-dimensional building model.
Instead of representing a wall primarily as lines in plan and elevation, BIM software can represent it as a model element with geometry, relationships and properties. Floors, roofs, columns, openings and other components can be modelled according to the agreed project scope.
This can be valuable where the design team already works in BIM, where three-dimensional coordination is important, or where the existing building needs to sit within a wider model-based workflow.
But a BIM model is still an interpretation of reality. Existing buildings are often irregular, distorted, altered or poorly aligned with idealised object behaviour. The modelling specification therefore needs to define how faithfully those irregularities should be represented and what information is actually required.
| Point Cloud | 2D CAD | BIM Model | |
|---|---|---|---|
| Representation | Dense measured 3D points | Interpreted 2D linework | Interpreted 3D objects |
| Typical formats | E57, RCP/RCS and others | DWG, PDF | RVT, IFC where scoped |
| Geometry retained | Large amount of visible measured surface information | Selected geometry required for drawings | Geometry selected and modelled to agreed specification |
| Human interpretation | Lower at raw-data level | Significant drafting interpretation | Significant modelling interpretation |
| Easy to print/read as drawings | No | Yes | Drawings can be generated from model views |
| Useful for additional later measurement | Strong, subject to coverage and accuracy | Limited to information drawn | Limited to information represented in the model |
| Design workflow | Reference / measurement / coordination | Conventional 2D design | Model-based 3D design and coordination |
| File size / computing demand | Can be substantial | Usually relatively efficient | Varies with model complexity |
| Best when | Teams need access to rich measured survey data | The project primarily needs plans, elevations and sections | A structured 3D model forms part of the project workflow |
None of these columns is a quality ranking. The appropriate deliverable depends on the work that follows the survey.
Not necessarily.
CAD and BIM are normally selective interpretations. The registered point cloud can preserve measured information that was outside the original drawing or modelling scope. Keeping access to it can therefore be useful if a later design question concerns an area that was captured but not originally drafted.
For example, a project may initially request floor plans and sections. Months later, an engineer may need to inspect the position of an overhead beam. If that beam is clearly represented in the retained point cloud, the survey data may still provide useful information even though it was not shown on the original floor plan.
This does not mean every client needs the complete point cloud. Large datasets require appropriate software, storage and management. The decision should be made deliberately as part of the survey handover.
For many measured-building projects, DWG/PDF drawings plus a registered point cloud can be a useful combination.
The drawings give the design team immediately usable plans, elevations and sections. The point cloud preserves richer measured information for reference and later questions.
If point-cloud delivery is required, agree the format before handover. E57, RCP/RCS, LAS and LAZ serve different workflows, and colour, scan structure, coordinates and other requirements should not be assumed from the file extension alone.
A point cloud also has independent value as survey data. Its preparation, cleaning, export, hosting or delivery may therefore form a specific part of the agreed commercial scope rather than being treated as an incidental by-product of the drawings.
BIM is most useful when the project will actually use the model.
A structured model may make sense when:
If the immediate requirement is simply a set of reliable plans, elevations and sections, commissioning a detailed BIM model can add work without necessarily adding equivalent project value.
The question is not "Can this survey be delivered as BIM?" It usually can. The better question is "What will the project do with the model that it cannot do efficiently with the point cloud and drawings?"
More model detail creates more production work, larger files and more decisions about how real-world geometry should be interpreted.
Existing buildings rarely behave like perfect design models. Walls can lean or vary in thickness. Historic fabric can be irregular. Structural elements may be partially obscured. Services can be densely layered.
A modelling brief should therefore identify which elements matter and how much geometric fidelity is needed. It is often more useful to model the right information consistently than to create a highly detailed model containing features the project will never use.
The same principle applies to CAD. Drawing every visible feature is not automatically better than producing a clear drawing containing the information appropriate to its intended scale and purpose.
A useful brief starts with the design task rather than a file format.
For an architectural refurbishment project, for example, the team might need:
Another project may be BIM-led from the outset and require a coordinated existing-condition Revit model with the point cloud retained as its measured reference.
If you are unsure, send the surveyor your design brief, existing drawings and intended software workflow. The deliverable package can then be scoped around what the team actually needs.
Geo Scan treats capture and deliverables as connected but separate decisions.
First, the site has to be measured with a methodology appropriate to the required accuracy, coverage and complexity. The registered survey data then provides the reference from which the required outputs are interpreted.
From that common measured base, we can provide point-cloud datasets, Scan to CAD drawings, Scan to BIM models or a combination of deliverables.
The best deliverable is not the one containing the most information. It is the one that gives the project team the information it needs in a form it can use confidently.