TL;DR
Building scanning for existing conditions verification pairs a capture method (terrestrial tripod, mobile SLAM, handheld, photogrammetry, or aerial) to a stated USIBD LOA target, then runs a documented workflow from control through QC. Zealot registers field data to ±5 mm and delivers model faces to ±10 mm at LOD 300 using a NavVis VLX3 (2.56M pts/sec) and Leica RTC360. This guide gives the decision table, the accuracy bands, the end-to-end workflow with real figures, the full deliverables manifest, and the acceptance-verification protocol an engineering team should require before sign-off.
Table of contents
- When you need a scan and when you don't
- Capture methods compared
- Setting the accuracy target
- The workflow, end to end
- Deliverables and formats
- Verifying what you receive
- Cost and schedule
- FAQ
When you need a scan and when you don't
A scan earns its cost when the risk of relying on record drawings exceeds the price of verified geometry — renovation, addition, tenant improvement, litigation, or any project where a wrong dimension moves a wall, a duct, or a column line. It is not justified for early feasibility studies where rough dimensions suffice.
| Project type | Scan recommended | Why |
|---|---|---|
| Renovation / gut-fit-out | Yes | Record drawings rarely match field conditions after decades of change orders |
| MEP coordination in occupied building | Yes | Clash risk between new and existing systems is high and expensive to fix after fabrication |
| Historic preservation | Yes | Irregular geometry cannot be estimated from drawings |
| Litigation / insurance dispute | Yes | Independent, defensible, timestamped record of conditions |
| Early feasibility / massing study | No | Approximate dimensions from existing drawings are adequate at this stage |
| Building already has a current, verified as-built set | No | Re-scanning duplicates existing accurate data |
| Interior-only paint/carpet refresh with no wall moves | No | No dimensional risk exists to mitigate |
The honest "don't" cases matter as much as the "do" cases: a scan adds cost and schedule, and a team that recommends scanning for every project regardless of risk is optimizing for its own revenue, not the client's outcome. See /services/building-3d-laser-scanning for scope-of-work guidance on framing this decision at proposal stage.
Capture methods compared
No single capture method is correct for every building. The right choice depends on square footage, occupancy, required accuracy class, and how much detail is needed at tie-in points versus open floor plates. Mixing methods on one project is normal and often the efficient answer.
| Method | Typical accuracy class | Throughput | Best use | Limitations |
|---|---|---|---|---|
| Terrestrial tripod (Leica RTC360) | USIBD LOA 10–30, survey-grade | Slower per station, highest fidelity per setup | Tie-in points, complex MEP, structural detail, control validation | Slow across large open floor plates; occlusion at height |
| Mobile SLAM (NavVis VLX3) | USIBD LOA 20–40 | 2.56M pts/sec while walking | Large floor plates, corridors, multi-story circulation | Slightly lower per-point precision than static tripod scans |
| Handheld SLAM | USIBD LOA 30–40 | Fast, tight spaces | Mechanical rooms, crawlspaces, stairwells | Accuracy drifts over very long unclosed loops |
| Photogrammetry | USIBD LOA 30–50 | Fast capture, slower processing | Facade documentation, textured surfaces, visual record | Weak on featureless or reflective surfaces; no direct dimensional control without ground truth |
| Aerial (drone) | USIBD LOA 30–50 | Very fast over large exteriors/roofs | Roof condition, site context, tall facades | Not usable indoors; airspace and safety restrictions apply |
As a working rule, mobile SLAM becomes the more efficient primary method above roughly 80,000–120,000 sq ft per floor, with tripod stations added selectively for tie-in accuracy and any area needing survey-grade detail. Below that threshold, tripod-only capture is often faster overall because setup time is amortized across less area. Zealot runs NavVis VLX3 as the primary instrument on qualifying projects and brings in the Leica RTC360 for control ties and high-detail zones.
Setting the accuracy target
Accuracy should be specified as a USIBD Level of Accuracy (LOA) band tied to the project's actual use case, not defaulted to the tightest available number. Each step up in LOA adds field time, processing time, and cost, so the target should be no tighter than the decision it supports.
| USIBD LOA | Represented accuracy | Measured accuracy | Typical use case |
|---|---|---|---|
| LOA 10 | Approximate, schematic | Not independently verified | Early planning, massing studies |
| LOA 20 | ± 15 mm class | Verified to stated tolerance by spot check | Conceptual coordination, space planning |
| LOA 30 | ± 9–15 mm class | Verified by sampled check measurements | General as-built documentation |
| LOA 40 | ± 5–9 mm class | Verified against surveyed control | MEP coordination, retrofit design |
| LOA 50 | ± 5 mm class or tighter, project-specific | Verified against surveyed control with full residual reporting | Structural verification, forensic/litigation, historic preservation |
"Represented Accuracy" is the tolerance a deliverable is drafted to; "Measured Accuracy" is the tolerance independently confirmed by check measurement against the real building. A model can be represented at LOA 40 and still fail if no measured verification was performed — the two figures are not interchangeable, and a contract should require both. Zealot's field registration runs to ±5 mm and model faces are delivered to ±10 mm at LOD 300, which sits at the LOA 40–50 range; specifying LOA 50 for a routine tenant improvement is typically overkill and adds cost without changing the design decision it informs.
The workflow, end to end
A complete existing-conditions scan runs seven stages from scoping to delivery, and each stage has a defined output and a defined failure mode. Skipping a stage — most often control or QC — is the most common cause of a deliverable that fails acceptance later.
| Stage | What happens | Typical duration | What you receive | What can go wrong |
|---|---|---|---|---|
| 1. Scoping | Define LOA target, deliverable formats, boundary, occupancy constraints | 1–3 days | Written scope of work with accuracy and format commitments | Undefined LOA leads to disputed acceptance later |
| 2. Control | Establish or tie into a campus/site control network with a total station or GNSS | 0.5–2 days | Control network report, closed to a stated tolerance | Weak control propagates error into every downstream cloud |
| 3. Field capture | Terrestrial, mobile SLAM, and/or handheld capture per the method plan | 1–5 days depending on square footage | Raw scan data, field log | Missed coverage in mechanical rooms or above ceilings |
| 4. Registration and QA | Cloud-to-cloud and cloud-to-control registration, residual review | 1–3 days | Registered point cloud, registration report | Unreviewed residuals hide localized drift |
| 5. Cleanup and segmentation | Noise removal, deduplication, splitting by discipline/floor | 1–2 days | Cleaned, segmented cloud | Over-aggressive cleanup deletes real detail |
| 6. Drafting / modeling | 2D CAD or Revit modeling to the specified LOD and centerline/face convention | 1–3 weeks depending on scope | CAD or BIM deliverable | Ambiguous LOD instructions cause rework |
| 7. QC and delivery | Independent check measurements, deviation review, sign-off | 2–5 days | QC report, final deliverable package | Skipping independent checks leaves errors undetected until construction |
Zealot's campus control network work has closed to ±6 mm on multi-building projects, and a documented hospital-wing project ran 36 QC checks at 4.2 mm mean error against a ±6 mm registered accuracy target. On a 220,000 sq ft plant, field verification found 143 pipe runs routed differently than the record PDFs showed — a finding only possible because registration and QC stages were run rigorously rather than skipped. See /services/as-built-documentation for how this workflow maps to specific as-built formats.
Deliverables and formats
A deliverables manifest should list every file, its extension, and the software that opens it — a scope that says "as-built drawings" without naming formats invites disputes at delivery. The underrepresented deliverables below are frequently omitted from provider proposals despite being routine to produce.
| Deliverable | File extension(s) | Opens in | Typical use |
|---|---|---|---|
| Registered point cloud | .e57, .rcp/.rcs | Recap, CloudCompare, Revit | Base data for modeling and coordination |
| 2D as-built drawings | .dwg, .pdf | AutoCAD, Bluebeam | Permit sets, construction reference |
| BIM model | .rvt, .ifc | Revit, Navisworks, ArchiCAD | Coordination, clash detection, FM handoff |
| Orthographic images | .tiff, .jpg | Any image viewer, CAD underlay | Facade and floor plan reference, quick visual QA |
| Floor flatness report | .pdf, .xlsx | Excel, PDF viewer | Slab tolerance verification (FF/FL or deviation-based) |
| Deviation analysis | .pdf, .rcp overlay | Recap, deviation-analysis software | As-built vs design comparison, renovation planning |
| BOMA lease plans | .dwg, .pdf | AutoCAD, PDF viewer | Leasing, area certification |
| MEP equipment families | .rfa | Revit | Facilities management, coordination models |
| QC report | PDF viewer | Documented proof of accuracy claims | |
| Panoramic/site imagery | .jpg, .e57 (embedded) | NavVis IVION, web viewer | Remote site walkthrough, verification reference |
Orthographic images, floor flatness reports, deviation analysis, BOMA lease plans, MEP equipment families, and the QC report itself are the deliverables most often left off a competitor's proposal, yet each is standard production output once the point cloud exists — the marginal cost of adding them is far lower than commissioning them separately later.
Verifying what you receive
An engineering team should not accept a deliverable on the provider's stated accuracy alone; acceptance should be based on an independent check-measurement protocol run against the delivered model or drawings, with defined sample size, pass criteria, and required report contents.
Acceptance-verification protocol:
- Confirm the registration report is included and states bundle adjustment, cloud-to-cloud RMS per station pair, maximum single-station residual, and control-network closure (target ±6 mm class or better, per #4 point-cloud-qa-scan-to-bim).
- Select an independent sample of check measurements — sized to project complexity, with 36 checks used on a documented hospital-wing project as a reference scale.
- Take each check measurement physically, independent of the scan data, and compare it to the same dimension in the delivered model or cloud.
- Require a mean error at or below the stated Measured Accuracy for the specified LOA band (Zealot's reference figure: 4.2 mm mean error against a ±6 mm target).
- Review a deviation map for any localized area exceeding twice the mean error, and flag it for re-check before sign-off.
- Confirm the QC report documents pass/fail status for every sampled check, not only a summary average.
- Withhold final acceptance until all flagged deviations are re-measured and resolved or formally accepted as known conditions.
This protocol should be written into the contract at scoping stage (see #10 revit-checks-before-hiring-scan-to-bim-services), not negotiated after a dispute arises.
Cost and schedule
Building scanning for existing conditions typically runs $0.05–$0.20 per sq ft, with the driver being required LOA, deliverable complexity (2D vs full BIM), occupancy constraints, and site access. A straightforward LOA 30 as-built on an unoccupied 100,000 sq ft warehouse sits at the low end; an LOA 50 structural verification with a full Revit model on an occupied hospital sits at the high end.
| Project scale | Field capture | Registration & QA | Modeling/drafting | Total turnaround |
|---|---|---|---|---|
| Under 20,000 sq ft, 2D deliverable | 1 day | 1–2 days | 3–5 days | 1–2 weeks |
| 20,000–100,000 sq ft, BIM deliverable | 2–4 days | 2–3 days | 2–3 weeks | 3–5 weeks |
| 100,000–220,000 sq ft, multi-discipline | 4–8 days | 3–5 days | 3–6 weeks | 5–8 weeks |
A staged turnaround is available on most projects: registered point cloud first (fastest), 2D drawings second, full BIM model last — letting design teams start early coordination before the complete model is finished. Call 614-210-3679 to scope turnaround against a specific building profile.
FAQ
Why record drawings are not enough
Record drawings represent the building as designed or as last documented, not as it stands today. Decades of tenant changes, field modifications, unpermitted alterations, and undocumented repairs accumulate between the last drawing revision and the present condition, and none of that drift is visible on a PDF. Verified geometry closes that gap with a defensible, timestamped dataset rather than an assumption. Structural, MEP, and architectural teams that design against unverified drawings inherit every one of those undocumented changes as unbudgeted risk once construction starts and a wall, duct, or beam is not where the drawing says it is.
Coordinating scanning with other trades and consultants
A scan is most valuable when its outputs are shared early with every consultant working from the same building, not siloed inside one discipline's model. Structural, MEP, and architectural teams pulling geometry from a single registered point cloud avoid the version-control problem of each discipline maintaining its own separately re-measured baseline, and clash detection performed against one common dataset is materially more reliable than clash detection performed against drawings assembled from different survey dates. Scoping should specify which disciplines receive native model access (.rvt, .ifc) versus reference-only formats (.pdf, .dwg), since access level affects both cost and how easily the data can be kept current as design progresses. See /industries/engineering for how this coordination plays out on multi-discipline capital projects, and /blog/what-is-reality-capture for the underlying capture-to-deliverable pipeline shared across disciplines.
Common mistakes that undermine a scanning program
Several recurring mistakes reduce the value of an otherwise well-executed scan. Specifying a deliverable format without specifying the LOA target leaves accuracy undefined and unenforceable. Accepting a stated accuracy figure without a registration report or independent check measurements means the figure is unverified. Treating LOD as if it implies dimensional accuracy leads teams to under-specify tolerance and discover the gap only after fabrication. Skipping control-network tie-in on multi-building campuses causes clouds from separate capture sessions to drift relative to one another even when each is internally accurate. And omitting a QC stage from the schedule — often cut first when a project runs late — removes the only check that catches a registration or modeling defect before it reaches the design team. Each of these is preventable by writing the accuracy target, verification protocol, and deliverables manifest into the scope of work before capture begins, as outlined in #4 point-cloud-qa-scan-to-bim and #10 revit-checks-before-hiring-scan-to-bim-services.
Working with occupied buildings
Scanning an occupied building adds scheduling and access constraints but does not change the underlying accuracy achievable with either terrestrial or mobile SLAM methods. Field crews typically work around occupancy by scheduling capture during off-hours, sequencing floor-by-floor to minimize disruption, and using mobile SLAM to move through corridors and common areas quickly rather than setting up a tripod in active workspace. Coordination with building management on access windows, elevator use, and any areas requiring escort is usually the larger scheduling variable, not the capture technology itself. A scope of work for an occupied building should state these access constraints explicitly alongside the accuracy target, since access limitations can extend field duration even when square footage is modest.