If you've ever tried to renovate a building with no reliable drawings, you already know the problem. Scan to BIM services exist precisely because the world is full of structures that were built before digital documentation, or whose paper drawings are wrong, outdated, or simply lost.
Here's how the process actually works and what separates a clean deliverable from a costly mess.
Why Existing Drawings Are Almost Never Enough
Architects and engineers tend to trust record drawings more than they should. A 1980s office block might have a set of CAD files from a 2003 refurbishment, but those files probably don't reflect what the contractor actually built. Walls shifted. Columns were added. Ductwork was rerouted.
Walking a site with a tape measure fixes this partially. 3D laser scanning fixes it comprehensively.
A terrestrial scanner a Leica RTC360, for instance can capture up to two million points per second, producing a dense spatial record of every surface in a room within millimetres of accuracy. That point cloud becomes the raw material for a Scan to BIM workflow.
The Scan to BIM Process, Stage by Stage
Stage 1: Site Scanning
A survey team sets up the scanner at multiple positions across the site. The exact number of scan positions depends on building complexity — a single-storey warehouse might need 20 setups; a multi-storey hospital wing might need 200 or more. Each scan captures a full spherical view of the surrounding environment, including walls, floors, ceilings, structural elements, and MEP components.
Targets (physical reference markers) are placed between positions so the individual scans can be aligned, or "registered," into a single unified model.
Stage 2: Point Cloud Registration
Back in the office, the raw scans are imported into registration software such as Leica Cyclone, FARO Scene, or Autodesk ReCap. The software stitches the individual scan positions together using the shared targets as reference points. The result is a single, georeferenced point cloud — typically in the range of tens to hundreds of gigabytes for a large site.
Colour is usually retained, mapped from the scanner's integrated camera. This makes it significantly easier for a BIM modeller to distinguish between, say, a concrete column and a steel pipe wrapped in insulation.
Stage 3: Point Cloud to BIM Modelling
This is where the raw data becomes useful. A BIM technician imports the registered point cloud into Autodesk Revit (or a similar platform) and begins modelling over it, creating actual BIM elements — walls, slabs, beams, columns, pipes, ducts, and so on — that correspond to what the scanner recorded.
The level of detail required here is defined in advance. In ISO 19650 terms, this maps to the Level of Information Need (LOIN). A structural engineer preparing for a load assessment needs different information than a mechanical contractor planning a new plant room installation. Define this early; retrofitting it later wastes everyone's time.
Stage 4: Quality Control and Clash Review
A finished as-built BIM model should be verified against the original point cloud before it leaves the production team. A common check is to measure the deviation between a modelled wall face and the corresponding cluster of scanned points — anything beyond ±5mm for a standard architectural model warrants review.
Clash detection follows. Even in as-built documentation, elements sometimes conflict with each other when properly modelled in 3D, revealing conditions the original site team simply worked around and never recorded.
Stage 5: Model Delivery
The final deliverable is typically a Revit (.rvt) file, accompanied by IFC exports for use in non-Revit environments, and a linked point cloud file for reference. Some clients also request 2D outputs — floor plans, sections, elevations — derived directly from the model.
What LOD Actually Means for Your Project
LOD — Level of Development — is one of the most misused terms in BIM conversations. Clients often ask for "LOD 300" without being certain what that includes for their specific use case.
For as-built BIM modeling on a renovation project, consider what each tier actually delivers:
- LOD 200 — Generic shapes, approximate sizes and locations. Useful for early feasibility.
- LOD 300 — Geometrically accurate elements, dimensionally correct and properly located. This is the standard for most renovation briefs.
- LOD 350 — Adds connection and interface information — how elements connect to adjacent systems. Necessary for MEP coordination.
- LOD 400 — Full fabrication-level detail. Rarely needed for as-built documentation unless you're handing off to a contractor for off-site fabrication.
Don't pay for LOD 400 when LOD 300 is what your project actually needs. And don't accept LOD 200 when your structural engineer is depending on accurate column positions.
Where Scan to BIM for Renovation Saves Real Money
A heritage hotel project in the UK — not a client we can name, but a situation we've encountered more than once — discovered during a Scan to BIM survey that a load-bearing wall the original drawings showed as 200mm thick was actually 350mm in one section and 180mm in another. The structural retrofit design had assumed uniformity. Catching that discrepancy before construction began saved a significant remediation cost.
That's the practical argument for 3D laser scanning to BIM on renovation work. Assumptions about existing conditions are expensive. Measured data is not.
Common Mistakes That Kill a Scan to BIM Project
Scanning without a clear BIM brief. The scanning team and the modelling team need to agree on scope, LOD, and coordinate system before the scanner touches the site. Misaligned expectations at this stage create rework that neither party budgets for.
Underestimating data management. A full building scan generates large files. Make sure your IT environment — or your outsourcing partner's environment — can handle the processing load without bottlenecks.
Ignoring occluded areas. Scanners work on line of sight. Spaces behind ductwork, inside ceiling voids, or beneath raised floors may not be captured. A good surveyor flags these gaps; a careless one delivers a point cloud with holes and says nothing.
Treating the point cloud as the deliverable. Point cloud data is not a BIM model. It's raw material. The intelligence in the model — element properties, system relationships, schedule data — has to be added by experienced BIM technicians. One without the other is half a job.
At Eminent BIM Services, we work with scan data supplied by our clients' survey teams, or coordinate directly with licensed surveyors on behalf of clients who don't have that resource in-house. Our modelling team works in Autodesk Revit across all disciplines — architectural, structural, and MEP — and we deliver models compliant with ISO 19650 information management requirements.
We've handled everything from single-storey retail units to multi-building industrial campuses. The process is the same; the complexity scales. What never changes is the need for a clearly defined scope before anyone picks up a scanner.
The One Question Worth Asking Before You Start
Before commissioning a Scan to BIM project, ask this: what decision will this model be used to support? Renovation design, structural assessment, MEP retrofit, space planning, facility management — each has different requirements at the model level.
Answer that question first, and the rest of the scope practically writes itself. Get it wrong, and you'll spend more correcting the model than you saved by scanning in the first place.
What's the existing building condition you're trying to document — and do you already have a survey partner on board?