A Matterport scan becomes a BIM model in four stages: capture, point cloud registration, dimensional QA, and model authoring in Revit or a comparable platform. For most commercial buildings, that path runs five to ten business days from scan day to a validated LOD 200–300 model, depending on building size and the level of detail specified.
This post is for architects, builders, and facility owners who have ordered a scan-to-BIM deliverable — or are deciding whether to — and want to know what actually happens in that pipeline, what drives the price, and where the process still needs a person, not just software.
What does “reality capture feeds BIM” actually mean?
It means the BIM model is built from measured, as-found conditions instead of assumptions. A Matterport scan captures millions of measured points across a space — walls, ceilings, mechanical runs, structural elements — and that point cloud becomes the dimensional foundation the BIM model is built on. Instead of a modeler estimating from old drawings or a single field visit, every wall, offset, and ceiling height in the model traces back to a measured point.
This matters most on buildings without reliable as-built drawings: older commercial buildings, buildings that have been renovated multiple times, or buildings where the original drawings don’t match what’s actually there. The scan replaces guesswork with a dimensional record.
What happens between the scan and a usable Revit model?
Four steps, in order:
- Capture. A Matterport scan of the space, walked in a single site visit. Scan time depends on square footage and how many rooms or levels need coverage.
- Point cloud registration. Individual scan positions are stitched into one continuous point cloud, accurate to within about 1% at typical range — the same precision standard used across P3D’s Matterport deliverables.
- Dimensional QA. Before modeling starts, the point cloud is checked against known reference dimensions — door widths, ceiling heights, structural bay spacing — to catch registration errors before they carry into the model.
- Model authoring. A modeler builds the BIM elements — walls, floors, ceilings, structure, and any specified MEP — inside the point cloud, at the level of detail the client specified. This is the step where a real person makes judgment calls: where a wall assembly transitions, how to model a soffit, what counts as a separate object versus a single family.
Steps one and two are largely automated. Steps three and four are where a scan-to-BIM deliverable earns its price — someone has to verify the data and make modeling decisions a script can’t make on its own.
What level of detail should you specify — and why does it change the price?
LOD (Level of Development) sets how much detail and reliability the model carries, and it’s the single biggest driver of scan-to-BIM cost. In practical terms:
- LOD 200 gives you approximate geometry — good for early space planning and renovation feasibility, not for construction bidding.
- LOD 300 gives you accurate, coordinated geometry suitable for construction documents and trade coordination.
- LOD 350 adds interface and connection detail between systems — used when trades need to coordinate clashes before they hit the field.
Specifying a higher LOD than the project needs is the most common way scan-to-BIM budgets run over. A feasibility study doesn’t need LOD 350. A full MEP coordination package usually does. The right move is to specify LOD per building system — for example, LOD 200 for architecture and structure, LOD 300 for MEP.
Where does automation stop and a modeler have to take over?
Software can register a point cloud and even auto-generate rough wall and floor surfaces. It cannot reliably tell the difference between a structural column and a duct chase, decide how to model an as-built condition that doesn’t match any standard family, or catch a scan gap caused by furniture blocking a wall. Every scan-to-BIM deliverable P3D produces goes through a architectural reviewer who checks the automated output against the actual space before it’s called final. That review step is what keeps a model usable instead of just fast, and it’s the reason P3D treats scan-to-BIM as a service with a person accountable for the output, not a fully automated conversion.
What does a scan-to-BIM deliverable actually cost, compared to a standard tour?
A standard Matterport tour — the walkable 3D model and floor plan — runs $0.10 to $0.25 per square foot depending on building size and scope, as detailed in P3D’s Matterport pricing breakdown. A scan-to-BIM deliverable adds the point cloud processing, QA, and modeling labor on top of that base capture cost, and the increase scales with LOD: a LOD 200 shell model adds the least; a fully coordinated LOD 350 package with MEP adds the most. Because pricing depends on building type, square footage, and specified LOD, P3D scopes every scan-to-BIM project individually rather than quoting a flat per-square-foot rate for BIM the way it does for a standard tour.
Building type also affects price, independent of square footage. An open warehouse floor plan with few interior walls models faster than an equivalent square footage of small offices, corridors, and mechanical rooms, because more individual elements mean more manual modeling time even at the same LOD. A project with unusual existing conditions — additions built in different eras, non-standard structural systems — also runs longer than a straightforward single-era building, since more of the modeling requires custom judgment rather than repeatable standard elements.
A note on equipment: Perspective 3D deploys different capture devices depending on the accuracy tolerance a project calls for. Matterport pro series devices handles spatial documentation, virtual tours, and general condition records. LiDAR is used when a project requires tighter measurement precision — for example, MEP measurements and clash detection. Device selection happens during scoping, based on what the deliverable actually demands
How does this play out on an actual project?
Take a mid-size architecture firm in Grand Rapids working on a renovation of a 1970s office building with no reliable as-built drawings. A same-day Matterport scan captures the existing conditions. The point cloud becomes the dimensional base for a LOD 300 Revit model of the architecture and structure, while MEP is modeled at LOD 200 since the mechanical contractor plans to field-verify runs before installation anyway. The design team gets a coordinated existing-conditions model to design against, without a field crew re-measuring a building that’s already been measured once, accurately.
Where does this approach break down?
Matterport Scan-to-BIM and LiDAR Scan-to-BIM works best when the goal is documenting existing conditions accurately and quickly. It has real limits. A scan can’t see inside a wall cavity, so anything hidden — knob-and-tube wiring, an abandoned duct run, structural framing behind finished ceilings — still needs a probe, a destructive investigation, or existing record drawings to confirm. Extremely tight or fully furnished spaces can leave scan gaps that a modeler has to flag and estimate around rather than measure directly. And a rush timeline that skips the QA step is a rush timeline that produces a model nobody should build from — the dimensional check exists because point clouds do occasionally register with small errors, and catching those before modeling starts is cheaper than catching them on site. This is also why P3D doesn’t offer same-day BIM modeling the way it does same-day Matterport scan delivery — the point cloud can be captured and registered quickly, but the modeling and review steps that turn it into something design and construction teams can rely on take the time they take.
How does building size affect the Matterport to BIM timeline?
Building size changes the modeling timeline more than it changes the scan time. A single Matterport scan can cover a 20,000-square-foot building in a single day, regardless of what happens after. But modeling that same building at LOD 300 across architecture, structure, and MEP takes meaningfully longer than modeling a 5,000-square-foot space at the same LOD, because every wall, door, and system component the modeler builds is manual work layered on top of the automated point cloud registration. As a rough guide: a small commercial space (under 5,000 square feet) at LOD 200-300 typically lands at the shorter end of the five-to-ten-business-day range; a large multi-story building at LOD 300 with MEP, or any project specifying LOD 350 coordination, tends to land at the longer end or beyond, depending on how many building systems are in scope. P3D confirms a specific timeline once square footage and LOD are known, rather than quoting a single number across every project size.
Who reviews the model before it’s called final?
Every scan-to-BIM deliverable goes through two layers of review before it’s marked final: an internal QA pass that checks the model against the point cloud and known reference dimensions, and a client-facing review where the architect or owner spot-checks the delivered model against the acceptance criteria set at order time. The internal pass catches modeling errors — a wall built at the wrong height, a door swing modeled backward, a family assigned the wrong type. The client review catches scope mismatches — a system modeled at a different LOD than expected, or a naming convention that doesn’t match the firm’s existing Revit standards. Both reviews matter, because a model that passes internal QA can still fail a client’s expectations if the original order wasn’t specific enough about what “done” looks like.
Does a Matterport scan replace a licensed as-built survey?
For most construction documentation purposes, yes — a Matterport-based point cloud meets or exceeds the accuracy of a traditional field measure. For work requiring a stamped survey (certain permitting or legal boundary situations), a licensed surveyor is still required regardless of scan data.
Can P3D model MEP systems from a Matterport scan?
Yes, to the extent MEP is visible during the scan. Exposed ductwork, visible piping, and accessible equipment can be modeled directly from the point cloud. Concealed MEP behind finished ceilings or walls requires existing drawings, an access panel, or field verification — the scan can’t see through building materials.
How long does a scan-to-BIM project take from start to finish?
Capture happens in a single site visit. Processing, QA, and modeling typically run five to ten business days after that, depending on building size and specified LOD. Larger buildings or higher LOD packages run longer; P3D provides a specific timeline once scope and LOD are confirmed.
What file formats does P3D deliver for a BIM project?
Revit (RVT) is the standard deliverable, along with the source point cloud (E57) so the client retains the raw measured data independent of the model. CAD (DWG) exports are available on request.
Do I need to specify LOD before the scan happens, or can I decide after?
LOD should be specified before modeling begins, since it determines how much time the modeler spends per building system. The scan itself captures the same point cloud regardless of LOD — the difference is in how much of that data gets modeled and at what reliability.
Related reading: Scan-to-BIM Workflows: What Architects Need to Know · Understanding LOD in BIM from Reality Capture · From Matterport Scan to Construction-Ready As-Built · One Scan. Five Deliverables.
Have a building that needs a scan-to-BIM deliverable? Call 616-312-3947 or visit perspective3-d.com/contact.