The Reality Capture Glossary: 60+ Terms Explained in Plain English

This glossary defines more than 60 terms used in 3D laser scanning, reality capture, and scan-to-BIM work, each in plain English and each short enough to quote in an RFP or a kickoff meeting. It is written for architects, engineers, contractors, facility managers, and owners who need to read a scan proposal, write a scan deliverable spec, or understand what they are actually buying. Terms are grouped by topic — capture technology, data and file formats, modeling and BIM, accuracy and QA, and project workflow — rather than alphabetically.

Capture technology and methods

• Reality capture — Reality capture is the umbrella term for technologies that record the physical world as measurable digital data, primarily 3D laser scanning and photogrammetry. The output is typically a point cloud or mesh that documents a building, site, or object exactly as it exists, which then feeds downstream deliverables like BIM models, CAD drawings, and floor plans.

• LiDAR — LiDAR (Light Detection and Ranging) is a sensing technology that measures distance by emitting laser pulses and timing how long each pulse takes to reflect off a surface and return. A LiDAR scanner repeats this measurement hundreds of thousands of times per second, producing a dense set of 3D coordinates. LiDAR is the core technology behind most professional building scanning, whether mounted on a tripod, worn by an operator, or flown on a drone.

• Terrestrial laser scanning (TLS) — Terrestrial laser scanning is ground-based LiDAR scanning performed from a stationary tripod. The scanner captures everything visible from one position, then is moved to the next setup until the whole space is covered, and the individual scans are stitched together in registration. TLS is the traditional gold standard for accuracy and is still preferred when survey-grade control or very tight tolerances are specified, at the cost of slower fieldwork.

• Mobile mapping — Mobile mapping (or mobile LiDAR scanning) captures data continuously while the scanner is in motion — worn on an operator's back, pushed on a cart, or mounted on a vehicle. Because there are no stationary setups, mobile mapping can cover large buildings roughly 10x faster than tripod scanning; a 50,000 sq ft building typically takes 4–6 hours to capture. Modern mobile systems such as the NavVis VLX3 achieve ±5mm accuracy, sufficient for most architectural and engineering uses.

• SLAM — SLAM (Simultaneous Localization and Mapping) is the algorithm that makes mobile scanning possible: it continuously calculates where the scanner is in space while simultaneously building the map of the environment around it. SLAM fuses LiDAR data with inertial measurements to keep positioning accurate as the operator walks, and its quality is the main driver of a mobile scanner's overall accuracy.

• IMU (inertial measurement unit) — An IMU is a sensor package of accelerometers and gyroscopes that tracks a scanner's motion and orientation. In mobile mapping systems, the IMU works alongside SLAM to maintain accurate positioning between laser measurements, especially in feature-poor spaces like long corridors where LiDAR alone can drift.

• Photogrammetry — Photogrammetry reconstructs 3D geometry from overlapping 2D photographs by triangulating common features across images. It excels at producing colorized meshes and orthophotos from drone imagery and costs less to capture than LiDAR, but it depends on lighting and surface texture and generally delivers lower, less predictable accuracy indoors. Many projects combine both: LiDAR for geometry, photogrammetry for imagery.

• Drone / UAV scanning — Drone (UAV) scanning captures roofs, facades, sites, and stockpiles from the air using either a LiDAR sensor or photogrammetry cameras mounted on an unmanned aircraft. It is the practical way to document areas that are unsafe or slow to reach from the ground, and it pairs naturally with interior scanning to produce a complete inside-and-out dataset. Commercial drone operations in the US require an FAA Part 107 certified pilot.

• Scan station / setup — A scan station (or setup) is one stationary position from which a tripod scanner captures data. Buildings are scanned from many overlapping stations — often dozens or hundreds — so that every surface is seen from at least one position. Station count is a major cost driver in terrestrial scanning, which is why mobile scanning, with no stations at all, is faster in large spaces.

• 360° panoramic imagery — 360° panoramic imagery is spherical photography captured during scanning, viewable as an immersive walkthrough similar to street-view navigation. Panoramas add visual context that raw point clouds lack — signage, finishes, conditions — and are commonly delivered alongside point clouds for virtual site visits in as-built documentation packages.

• Intensity — Intensity is a value recorded with each LiDAR point describing how strongly the laser pulse reflected off the surface. Intensity data makes grayscale point clouds readable even without color photography — dark and light materials, painted markings, and surface changes all show up — and it is useful in spaces too dark for cameras.

• Time of flight — Time of flight is the distance-measurement principle used by most building-scale LiDAR scanners: distance equals the time a laser pulse takes to travel to a surface and back, multiplied by the speed of light. The alternative, phase-shift measurement, compares waveforms instead of timing pulses; both principles deliver millimeter-class range accuracy at building scale.

Point cloud data and file formats

• Point cloud — A point cloud is the raw output of 3D laser scanning: millions or billions of individual points, each with an X, Y, Z coordinate and often color and intensity values, that together form a measurable 3D picture of a space. A point cloud is a snapshot of reality, not an intelligent model — you can measure it, view it, and model from it, but its points carry no knowledge of what is a wall versus a pipe.

• Point density — Point density (or resolution) describes how closely spaced the points in a point cloud are, often expressed as average point spacing in millimeters. Higher density captures finer detail — ornate trim, small pipe — at the cost of larger files and longer processing. A scan spec should state the required density because "we scanned it" says nothing about how much detail was captured.

• Registration — Registration is the process of aligning multiple individual scans into one seamless point cloud in a shared coordinate system. Software matches overlapping geometry or targets between scans and computes the best fit, reporting a registration error that quantifies alignment quality. Poor registration is the most common source of bad scan data, which is why a QA report documenting registration results is a standard professional deliverable.

• Colorized point cloud — A colorized point cloud has RGB color values mapped onto each point from photographs captured during scanning, so the data looks like a photorealistic 3D image rather than a monochrome one. Colorization makes clouds far easier for non-specialists to interpret and helps modelers distinguish materials and systems during scan-to-BIM work.

• Noise — Noise refers to erroneous or unwanted points in a point cloud — reflections from glass and shiny surfaces, points captured on moving people or vehicles, and stray measurements in dust or fog. Professional processing includes noise filtering and cleanup so the delivered cloud contains real, stationary building surfaces.

• Decimation — Decimation is the deliberate thinning of a point cloud to a lower density to reduce file size and improve software performance. A decimated cloud (for example, thinned to 10mm point spacing) is often delivered for everyday design use alongside the full-density archive, since most workflows do not need every captured point.

• Mesh — A mesh is a continuous 3D surface built by connecting points into triangles, converting a point cloud or photogrammetry dataset into a solid-looking model. Meshes are the standard format for visualization, 3D printing, game engines, and volume calculations, whereas point clouds remain the standard reference for measurement and BIM modeling.

• Gaussian splatting (splats) — Gaussian splatting is a newer rendering technique that represents a captured scene as millions of small, soft, colored ellipsoids ("splats") rather than points or triangles, producing strikingly photorealistic, smoothly navigable 3D scenes from photo or scan data. Splats are excellent for visualization and stakeholder walkthroughs but are not a measurement or modeling format — they complement, rather than replace, a survey-grade point cloud.

• Orthophoto — An orthophoto is an aerial or elevation photograph that has been geometrically corrected so it has a uniform scale, like a map — you can measure distances directly on it. Orthophotos generated from drone photogrammetry are common site and roof deliverables, and true-to-scale facade orthophotos support elevation drawings and condition assessment.

• E57 — E57 is the vendor-neutral, ASTM-standardized point cloud file format, and the safest choice when data must move between different software ecosystems. An E57 file stores points, color, intensity, and scanner positions in one portable package, which is why it is the archival and exchange format most scan deliverable specs require. A fuller comparison is in our guide to point cloud file formats.

• RCP — RCP is Autodesk ReCap's project file format and the standard way to bring point clouds into Revit, AutoCAD, and Navisworks. An RCP file is actually an index that references one or more RCS scan files, so the RCP and its supporting RCS files must be delivered and stored together.

• RCS — RCS is Autodesk ReCap's individual scan data file — the container that actually holds the points an RCP project references. Recipients working in the Autodesk ecosystem need both: the RCP project plus every linked RCS file.

• LAS — LAS is a public point cloud format developed for airborne LiDAR and widely used in surveying, civil, and GIS workflows. LAS carries rich per-point attributes (classification, return number, GPS time) and is the expected format for drone LiDAR and topographic datasets.

• LAZ — LAZ is losslessly compressed LAS — identical data at roughly 5–10× smaller file size. LAZ is the practical choice for transferring or archiving large aerial datasets, and most GIS and survey software reads it natively.

• PTS — PTS is a simple plain-text point cloud format listing one point per line with its coordinates, intensity, and color. PTS files are easy for almost any software to parse but are enormous and slow compared to binary formats, so they are best treated as a lowest-common-denominator exchange option rather than a primary deliverable.

• COPC — COPC (Cloud Optimized Point Cloud) is a modern LAZ-based format organized so that software can stream just the portion of a huge point cloud it needs directly from web storage, without downloading the whole file. COPC is doing for point clouds what cloud-optimized formats did for satellite imagery, and it is increasingly used to serve large datasets in the browser.

• DWG — DWG is AutoCAD's native drawing format and the standard deliverable for 2D documentation extracted from scan data — floor plans, elevations, and sections produced through scan-to-CAD workflows. DWG files carry vector linework, layers, and dimensions rather than points.

Modeling and BIM

• BIM (Building Information Modeling) — BIM is the practice of representing a building as an intelligent 3D model whose elements know what they are — a wall object carries its type, thickness, and materials, not just its shape. BIM models support design, coordination, quantity takeoff, and facility management in ways flat drawings cannot, and they are the primary deliverable of scan-to-BIM projects.

• Revit — Revit is Autodesk's BIM authoring platform and the de facto industry standard for architectural, structural, and MEP modeling in North America. Most scan-to-BIM deliverables are native Revit models (RVT files) built over a linked point cloud, so the model drops directly into the design team's existing workflow.

• Scan-to-BIM — Scan-to-BIM is the process of converting laser scan data into an intelligent BIM model: the building is scanned, the registered point cloud is linked into modeling software, and a technician models walls, floors, structure, and MEP systems over the points. The result is an accurate as-built model of existing conditions, typically delivered in Revit at a specified LOD and used for renovation design, coordination, and facility management.

• Scan-to-CAD — Scan-to-CAD converts laser scan data into traditional 2D CAD documentation — floor plans, elevations, sections, and details, usually in DWG format — rather than an intelligent 3D model. It is the right scope when a project needs accurate drawings but not a BIM model, and it typically costs less than full scan-to-BIM.

• LOD (Level of Development) — LOD is the industry scale (defined by the BIMForum LOD Specification) describing how developed and reliable each element in a BIM model is, from LOD 100 conceptual placeholders to LOD 500 field-verified records. LOD is the single most important term to specify in a scan-to-BIM contract because it drives modeling effort and cost more than any other variable; see our comparison of LOD 200 vs 300 vs 400 for how it changes a Revit model.

• LOD 100 — LOD 100 is conceptual: elements are represented as symbols or rough massing that indicates something exists, without reliable size, shape, or location. In scan-to-BIM work LOD 100 is rare, appearing mainly in feasibility and early planning models.

• LOD 200 — LOD 200 elements are modeled as generic placeholders with approximate size, shape, and location. An LOD 200 as-built model captures overall building geometry — walls, floors, roof, major equipment — and suits space planning, marketing, and early design phases where approximate is acceptable.

• LOD 300 — LOD 300 elements are modeled as specific objects with accurate size, shape, location, and orientation, making the model reliable for measurement and construction documentation. LOD 300 is the most commonly specified level for scan-to-BIM as-builts, and it is the level most renovation design teams actually need.

• LOD 350 — LOD 350 adds the interfaces between building systems to LOD 300: hangers, connections, clearances, and how elements meet other elements. LOD 350 is the standard target for MEP coordination and clash detection, where the space between systems matters as much as the systems themselves.

• LOD 400 — LOD 400 elements carry fabrication-level detail — sizing, connections, and assembly information sufficient to manufacture and install from the model. LOD 400 is specified when the model will drive prefabrication or spool drawings, and modeling to this level from scan data costs significantly more than LOD 300.

• LOD 500 — LOD 500 means elements are field-verified representations of what was actually installed — a record-model designation about verification rather than extra geometric detail. A scan-based as-built model is a natural path to LOD 500, since the point cloud itself is the field verification.

• IFC — IFC (Industry Foundation Classes) is the open, vendor-neutral BIM exchange format maintained by buildingSMART. An IFC export lets a Revit model be opened in ArchiCAD, coordination tools, and FM software without the original authoring platform, and public projects increasingly require IFC deliverables alongside native files.

• COBie — COBie (Construction Operations Building information exchange) is a structured data standard — effectively a rigorously formatted spreadsheet — for handing over equipment, space, and asset information from a BIM model to a facility management system. COBie strips the geometry and keeps the data an owner's maintenance team actually uses: what the assets are, where they are, and how to service them.

• Digital twin — A digital twin is a digital replica of a physical building or asset that stays connected to it over time, often combining a scan-based 3D model with live or regularly updated data such as sensors, maintenance records, and occupancy. In practice, most building digital twins start life as a laser scan and BIM model; the "twin" label is earned when the model is kept current and used operationally rather than archived.

• Clash detection — Clash detection is the automated checking of a coordinated model for physical conflicts — a duct routed through a beam, a pipe occupying the same space as conduit — before anything is built. Running clash detection against a model built from scan data means new design work is checked against real installed conditions, catching conflicts while they are still a software fix instead of a field problem.

• MEP coordination — MEP coordination is the process of arranging mechanical, electrical, and plumbing systems so they all fit within the available space along with structure and finishes, typically in a shared BIM environment. Renovation-phase MEP coordination depends on accurate existing-conditions scans, because the congested reality above a ceiling rarely matches decades-old drawings.

• VDC (Virtual Design and Construction) — VDC is the broader construction-management practice of building and rehearsing a project digitally — models, schedules, logistics, and coordination — before executing it in the field. BIM is VDC's core tool; reality capture supplies VDC teams with ground truth about existing conditions and installed work.

• Navisworks — Navisworks is Autodesk's project review and coordination platform, used to combine models and point clouds from many sources into one federated model for clash detection, 4D scheduling, and constructability review. Scan deliverables destined for coordination are commonly requested in Navisworks-compatible formats (NWC/NWD alongside RCP).

• Prefab / spool drawings — Spool drawings are fabrication-level drawings of pipe or duct assemblies that are manufactured off-site ("prefab") and installed as complete units. Prefabrication depends on trusting field dimensions, which is why MEP contractors scan existing conditions first — a ±5mm point cloud lets spools be fabricated to fit the first time instead of being field-modified.

Accuracy, QA, and standards

• Accuracy vs. precision — Accuracy is how close a measurement is to the true value; precision is how repeatable measurements are relative to each other. A dataset can be precise but inaccurate — internally consistent yet shifted from reality — which is why professional scanning verifies results against independent control rather than trusting instrument spec sheets alone. When a scanner is rated "±5mm," that figure describes the expected error band of its measurements.

• Absolute vs. relative accuracy — Relative accuracy describes how correct measurements are within the dataset (the length of a wall, point to point); absolute accuracy describes how correctly the whole dataset is positioned in a real-world coordinate system. A scan can have excellent relative accuracy while its absolute position is arbitrary — which is fine for a renovation model and unacceptable for work that must tie to survey or property coordinates.

• Survey grade — Survey grade describes measurement quality suitable for professional land-surveying tolerances, generally meaning centimeter-level or better absolute accuracy tied to established survey control. In building documentation the term is often used more loosely to mean millimeter-class relative accuracy (±5mm or better); a well-written scan spec avoids the ambiguity by stating the actual tolerance and whether survey control ties are required. Note that boundary and legal surveys remain the domain of licensed surveyors regardless of scan quality.

• LOA (Level of Accuracy) — LOA is the USIBD's standard scale for specifying and reporting measurement accuracy in building documentation, with bands from LOA10 (loosest, ±5cm and beyond) to LOA50 (tightest, ±1mm), each defined at the 95% confidence level. LOA answers "how accurate is the data," while LOD answers "how developed is the model" — a complete scan-to-BIM spec states both, because a highly detailed model built from sloppy data is still sloppy.

• USIBD — The USIBD (U.S. Institute of Building Documentation) is the industry body that develops standards and best practices for building documentation, best known for its Level of Accuracy (LOA) Specification and C-PAD documentation guides. Referencing USIBD LOA bands in a contract gives both buyer and provider an objective, measurable definition of acceptable accuracy.

• Georeferencing — Georeferencing is the process of placing scan data into a real-world coordinate system — such as a State Plane coordinate system with a known elevation datum — instead of an arbitrary local origin. Georeferencing is what allows a building scan to align with civil drawings, GIS data, property surveys, and scans captured years later.

• Control points — Control points are precisely surveyed reference points, often marked with checkerboard or spherical targets, used to anchor and verify scan data. Registering scans to control points constrains error accumulation across large sites and provides the independent check that turns claimed accuracy into demonstrated accuracy.

• Benchmark — A benchmark is an established reference point of known elevation (and often position) used as the basis for measurement on a site. Tying scan data to project benchmarks keeps scan-derived elevations consistent with the survey and construction documents everyone else on the project is using.

• Total station — A total station is the surveyor's precision instrument for measuring exact angles and distances to individual points, used in scanning workflows to establish control points and verify scan accuracy. Scanners capture millions of points quickly; a total station measures a few points extremely accurately — the two are complementary.

• GNSS / RTK — GNSS (Global Navigation Satellite System, including GPS) provides satellite positioning, and RTK (Real-Time Kinematic) is the correction technique that sharpens it from meters to centimeters. RTK GNSS is standard for georeferencing drone flights and outdoor scan control, but it does not work indoors — which is why interior scans are tied to coordinates through surveyed control points instead.

• Registration error (RMS) — Registration error is the reported misalignment between overlapping scans after registration, commonly summarized as an RMS (root mean square) value in millimeters. It is the headline quality metric in a scan QA report: a low RMS against independent checks is evidence the delivered cloud is internally sound.

• QA/QC report — A QA/QC report is the document a scanning provider delivers to substantiate data quality, typically covering registration results, control-point residuals, coverage verification, and achieved accuracy versus the specified LOA. A provider unwilling to deliver a QA report is asking you to take accuracy on faith; a good deliverable spec makes the report mandatory.

• Deviation analysis — Deviation analysis compares scan data against a design model or a reference geometry and maps the differences, usually as a color-coded heat map. It is used to verify that constructed work matches design (slab pours, steel placement, facade panels), to detect deformation over time, and to quantify how far existing conditions stray from plumb, level, and straight.

• Floor flatness (FF/FL) — Floor flatness (FF) and floor levelness (FL) are the ASTM E1155 numbers specifying how smooth and how level a concrete slab must be. Laser scanning captures an entire slab surface rather than sample points, producing complete FF/FL heat maps that show exactly where a floor is out of tolerance — critical for high-bay warehouses, robotics floors, and equipment installations.

• Tolerance — Tolerance is the amount of deviation from a specified dimension or position that is acceptable. In reality capture contracts, tolerance appears in two places: the accuracy tolerance of the data itself (the LOA), and the modeling tolerance describing how closely BIM elements must follow the point cloud (commonly stated as a maximum offset such as 10–15mm for as-built modeling). ZEALOT's standard: model faces within ±10mm of the registered cloud for LOD 300 elements, ±6mm for LOD 400.

Project and workflow terms

• As-built — As-built documentation records a building as it actually exists — what was really constructed, including every undocumented modification since — as opposed to design drawings showing what was intended. Laser scanning is the modern standard for producing as-built documentation because it captures actual conditions comprehensively instead of sampling them with a tape measure.

• Existing conditions — Existing conditions is the AEC term for the current physical state of a building or site at the start of a project — geometry, structure, systems, and features as they stand today. Documenting existing conditions accurately is the foundation of renovation and retrofit work; the phrase is effectively the "before" that as-built documentation captures.

• Scan planning — Scan planning is the pre-fieldwork process of deciding how a site will be captured: scan coverage and density, equipment choice, control strategy, access and safety logistics, and sequencing around occupants or operations. Good scan planning is why professional crews capture a building completely in one mobilization instead of returning for missed areas.

• Scan spec (scanning specification) — A scan spec is the written document defining exactly what a scanning project must deliver: coverage, point density, accuracy (LOA), file formats, coordinate system, modeling scope and LOD, QA requirements, and timeline. A clear spec is the single best protection against disputes over deliverables; our guide to writing a scan deliverable spec and scan-to-BIM scope-of-work template cover the details.

• Deliverable — A deliverable is any finished product handed over at the end of a scanning project: registered point clouds, Revit models, CAD drawings, floor plans, panoramic imagery, orthophotos, reports. Deliverables — not the scanning itself — are what a buyer is actually purchasing, which is why specs and proposals should be written around them; point cloud deliverables typically follow scanning by 3–5 business days, and BIM models by 2–3 weeks depending on scope.

• Mobilization — Mobilization is the effort and cost of getting a crew and equipment to a site and ready to work — travel, access coordination, safety training, badging. Mobilization is why small projects carry minimum fees and why capturing everything in one visit is cheaper than returning for missed scope.

• Facility management (FM) data — FM data is the operational information a building owner's team needs to run and maintain a facility — accurate floor plans, room areas, asset locations, equipment records — often extracted from scan data and BIM models and loaded into CMMS or IWMS platforms. Scan-based FM data replaces the folder of outdated drawings most facilities actually operate from.

• Adaptive reuse — Adaptive reuse is the conversion of an existing building to a new use — a warehouse into loft apartments, a church into offices, a school into housing. Because these buildings typically have missing or unreliable drawings and construction that deviates from them, adaptive reuse projects lean heavily on laser scanning to establish trustworthy existing conditions before design begins; see our guide to scan-to-BIM for adaptive reuse.

• Historic preservation documentation (HABS/HAER/HALS) — HABS (Historic American Buildings Survey), HAER (Historic American Engineering Record), and HALS (Historic American Landscapes Survey) are the National Park Service's standards for documenting historically significant structures, engineering works, and landscapes. Laser scanning is widely accepted in this work — point clouds capture ornament and conditions at millimeter resolution and support the measured drawings these programs require; more in our post on 3D scanning for historic preservation.

• Stockpile volumetrics — Stockpile volumetrics is the use of 3D scanning or drone capture to calculate the volume of bulk materials — aggregate, soil, coal, grain — by modeling the pile's actual surface. Scan-based volumetric measurement is typically accurate within 1–2% of true volume, outperforming traditional sparse-point surveys on irregular piles and providing defensible numbers for inventory and payment.

• Cut / fill — Cut and fill are the earthwork quantities on a site: cut is material that must be removed (excavated) and fill is material that must be added to reach design grade. Comparing a scanned existing surface against the design surface yields precise cut/fill volumes and maps, used for earthwork estimating, progress payment, and verification of contractor quantities.

• AEC — AEC stands for Architecture, Engineering, and Construction — the industry shorthand for the design and building sector that reality capture primarily serves. You will also see AECO, which adds Owners/Operations to reflect the facility management side of the building lifecycle.