3D Scanner

What is a 3D Scanner

A 3D scanner is a non-contact (or hybrid contact/non-contact) coordinate-measuring instrument that digitizes the geometry of a physical object into a point cloud or polygon mesh. Rather than reading a single dimension the way a caliper or a tactile probe does, a 3D scanner captures the full surface of a part as hundreds of thousands or millions of three-dimensional points, which downstream software converts into a triangulated mesh and, for reverse engineering, into NURBS surfaces. In the SpecForge taxonomy it sits under Test & Measurement › Vision & Imaging, and this is a single product TYPE page: the related links at the foot of the guide point to sibling categories such as Industrial Camera, Machine Vision System, Confocal Displacement Sensor and Laser Profile Scanner, not to sub-types of 3D scanner.

Industrial 3D scanners overlap with the optical-3D coordinate-measuring-system (CMS) family governed by ISO 10360-13. That overlap matters for procurement: a 3D scanner used for inspection is, in standards terms, a coordinate measuring system, and its stated accuracy should be defensible against the same acceptance and reverification framework that applies to a bridge or arm CMM. The instrument can be handheld and wireless, mounted on a tripod over a rotary table, fitted to an articulated arm, or set on a survey tripod to capture an entire building. What unites the family is the recovery of depth from optics rather than from mechanical contact.

Functionally, every 3D scanner solves the same problem in three stages: project or emit light onto a surface, image how that light is deformed or delayed by the surface, and compute three-dimensional coordinates from the geometry or timing of the returned signal. The differences between scanner types are differences in which optical principle performs that depth recovery — triangulation from a known baseline, distortion of a projected pattern, the round-trip time of a light pulse, the phase of a modulated beam, or the tracking of a retroreflector. Those principles, and the form factors built around them, are the subject of the next two chapters.

Photo: Kelly No.715, CC BY-SA 4.0, via Wikimedia Commons

Four engineering metrics dominate 3D scanner selection: accuracy (single-point or probing error), volumetric accuracy (which grows with measured length), measurement and mesh resolution, and the surface types the optics can handle. These determine whether a scanner can verify a given tolerance, how large a part it can digitize without accumulated error, and whether it copes with dark, shiny or translucent materials. The typical applications follow directly from those metrics: quality control and first-article inspection, reverse engineering, GD&T deviation analysis, rapid-prototyping and 3D-print preparation, tooling and die check, aerospace and automotive assembly metrology, medical and orthotics, forensics, and for the long-range units, architecture/engineering/construction (AEC), plant documentation and cultural-heritage preservation.

Scanner Types and Form Factors

3D scanners are first grouped by the optical principle they use to recover depth, and then by the physical form factor built around that principle. The principle determines accuracy, range and surface tolerance; the form factor determines where and how the instrument is used. The table below summarizes the six operating principles before the form factors are described.

Laser triangulation projects a laser line (or multiple parallel lines, or a cross) onto the surface; one or two cameras mounted at a known fixed angle — the baseline — image the deformed line, and depth is computed from the triangle formed by the laser source, the surface point and the camera. It is the workhorse for industrial parts, dark and reflective surfaces, and high-accuracy handheld metrology. Structured-light / fringe projection projects a sequence of patterns (sinusoidal fringes, Gray codes) over an area and observes their distortion with one or two cameras, triangulating every pixel simultaneously to capture a whole field of millions of points in one shot. Time-of-Flight measures the round-trip time of an emitted light pulse and is used in long-range terrestrial laser scanners (TLS) for buildings, plants and surveying, with lower point accuracy (millimeters to centimeters) but very long range. Phase-shift derives range from the phase difference of a continuous modulated beam, sits in the mid-range with higher accuracy than pure ToF, and is common in terrestrial and architectural scanners. A laser tracker follows a spherically mounted retroreflector (SMR) over distances up to tens of meters, combining interferometric or absolute-distance-meter (ADM) ranging with two angular encoders for large-volume metrology. Photogrammetry reconstructs geometry in software from many overlapping 2D photographs and is often paired with optical scanners to control accumulated error over large areas.

On top of these principles the market builds several form factors. Handheld / portable units are the most common in B2B metrology: wireless or tethered, they self-position by tracking reflective targets or surface geometry. Stationary / tripod structured-light systems combine a fixed sensor with a rotary table for the highest small-part accuracy. A portable CMM arm with a scan head — such as the Hexagon Absolute Arm or the FARO Quantum — combines tactile probing and laser scanning on one articulated arm. Desktop units serve entry, dental and jewelry work; terrestrial / long-range units serve AEC and plant documentation; and laser trackers serve aerospace and large-assembly metrology.

The practical consequence is that the same nominal task — "scan this part" — maps to very different instruments depending on size and surface. A 50 mm machined bracket with a matte finish is a stationary structured-light job; a 4 m casting with dark, reflective patches is a handheld blue-laser job referenced with targets; a 30 m aircraft jig is a laser-tracker job. Selecting on principle and form factor before comparing datasheets prevents the most expensive procurement mistakes.

Operating Technologies

This chapter goes deeper into the two technologies that dominate industrial metrology — laser triangulation and structured-light fringe projection — and explains where the long-range and large-volume principles fit. The distinction is not academic: it decides whether a scanner can see a chrome surface, how fast it captures, and how large a part it can measure before accumulated error becomes the limiting factor.

Laser triangulation remains the standard for high-accuracy handheld metrology because the triangle geometry is robust and the laser can be made bright enough to read difficult surfaces; the same triangulation principle underlies the single-point laser displacement sensor used for in-line gauging. The decisive recent change is the shift from red to blue laser: a shorter wavelength of around 450 nm is now standard on metrology units because it produces less speckle noise than red laser, capturing finer detail and coping better with dark and shiny surfaces. Modern handhelds project many parallel blue laser lines at once — for example 22 lines on the Creaform HandySCAN BLACK Elite or 30 lines on the HandySCAN BLACK+ Elite — to raise throughput without sacrificing the triangulation accuracy that the fixed baseline provides.

Structured-light / fringe projection trades some surface robustness for speed and field density. Because it triangulates every camera pixel from a projected pattern, it captures millions of points in a single shot, giving high resolution and accuracy in controlled lighting on matte surfaces. Its weaknesses are inherent to projected light: it is sensitive to ambient light, mitigated by a narrow-band blue LED source, and to shiny or translucent surfaces. Stereo-camera fringe systems — for example the ZEISS/GOM ATOS family — use a "Triple Scan" principle, combining the two cameras and the projector to validate each point and so handle reflective parts and recesses that defeat a single-camera system.

For everything larger than a workbench part, the principle changes again. Time-of-Flight terrestrial scanners measure the round-trip time of a light pulse to digitize buildings and plants at ranges of tens to hundreds of meters, accepting point accuracy in the millimeter-to-centimeter range as the cost of that reach. Phase-shift scanners improve on pure ToF accuracy in the mid-range and are common in terrestrial and architectural work, where they are often deployed alongside a surveying total station or theodolite that fixes the control network the scan registers to. Laser trackers occupy the large-volume metrology niche, following a spherically mounted retroreflector over tens of meters with interferometric or ADM ranging plus two angular encoders, which is how aerospace and large-assembly teams hold tight tolerances over very large work envelopes.

Across all of these, photogrammetry is the connective tissue rather than a competing scanner: by reconstructing geometry from many overlapping photographs it provides a globally accurate frame of reference into which the dense local scans are fitted, which is how large-part scanning keeps its per-meter error in check. In practice a high-end large-part job is rarely one technology — it is blue-laser or fringe scanning for local detail, targets for self-referencing, and photogrammetry to bound the accumulated error over the whole object.

Materials, Media and Surface Handling

3D scanners image surfaces optically, so surface optical properties dominate data quality. Unlike a tactile CMM, which is largely indifferent to color and gloss, an optical scanner depends on enough light returning cleanly to the camera, a constraint it shares with other optical-metrology instruments such as the vision measuring machine and the optical comparator. Dark, glossy or chrome, translucent or transparent surfaces scatter or absorb light and add noise or dropouts, and a part that scans perfectly in matte primer can become unmeasurable once it is anodized, polished or clear-coated. Surface, not material chemistry, is the variable that decides whether a scan succeeds.

The mitigation hierarchy runs from "change the light" to "change the part." First, blue laser or blue LED sources reduce speckle and reject ambient light better than red sources, which is why metrology units have standardized on roughly 450 nm blue. Second, HDR or multi-exposure capture lets the sensor record both the dark and the bright regions of a high-contrast surface in one merged scan. Third, Triple-Scan stereo geometry uses two cameras plus the projector to validate each point, recovering reflective parts and recesses that a single camera cannot. Only as a last resort does the part itself change: a thin matting spray, such as a titanium-dioxide coating or a vanishing developer powder, gives a temporary diffuse surface, at the cost of an added film thickness and a cleaning step.

Positioning targets are the other half of media handling. Retroreflective dots placed on or around the part let a scanner self-reference frame to frame and bound the accumulated error that otherwise grows across a large object; they are often combined with photogrammetry, which fixes the global frame while the scanner fills in local detail. Some structured-light systems, such as the Artec Eva, are positioned to scan many parts without targets or powder at all, trading the tightest tolerance for fast, general-purpose capture — a reminder that the right surface strategy depends on the accuracy class of the job.

Finally, the operating environment is part of the media problem. Thermal stability and vibration affect large-volume accuracy directly: a temperature swing changes the dimensions of both the part and the reference frame, and vibration blurs the projected pattern or laser line. For shop-floor metrology that means controlling temperature where possible, isolating the setup from vibration, and treating ambient light as something to be excluded rather than tolerated. The cleanest spec on a datasheet is only achievable inside the environment the datasheet assumed.

Key Specification Parameters

Reading a 3D scanner datasheet is a fundamental skill for purchasing engineers, and the trap is that the headline number — often a single accuracy figure — hides several independent specs. The parameters that actually drive selection are accuracy and volumetric accuracy, measurement and mesh resolution, measurement rate, point distance and field of view for structured light, working distance, part size range, light source, depth of field, weight, and optional texture or color capture. The table below gives correct units and typical industrial ranges for each.

Accuracy versus volumetric accuracy is the distinction most often missed. Accuracy is a single-point or probing error — how close one measured point is to the true dimension — and metrology-grade handhelds reach about 0.025 mm. Volumetric accuracy is stated as a base plus a length-dependent term, for example 0.020 mm + 0.040 mm/m, because error grows with the measured length; the best units reach 0.020 mm + 0.015 mm/m. For a small part the base term dominates and the two figures are close; for a large part the per-meter term dominates and becomes the figure that actually decides the result.

Resolution and measurement rate describe density and throughput, not closeness to truth. Measurement resolution (0.025 to 0.10 mm) is the point spacing the sensor can resolve, while mesh resolution (roughly 0.05 to 0.50 mm) is the density of the final triangulated mesh. Measurement rate ranges from 0.5 to 2.0 million measurements per second on handheld laser units, while structured-light systems capture up to about 12 million points in a single scan. For structured light, point distance (0.03 to 0.15 mm) and field of view / measuring volume (from roughly 100×70 up to 500×370 mm, on interchangeable volumes) trade against one another: a larger volume covers more part per shot but spreads the points farther apart.

The remaining parameters are practical constraints. Working / stand-off distance runs 200 to 750 mm on handhelds and around 490 mm on a typical structured-light head, which sets how close the operator must work. Part size range spans 0.05 to 4 m for handhelds and up to tens of meters for TLS and laser trackers. Depth of field of tens of millimeters up to about 550 mm on the latest handhelds sets how much the operator can move toward or away from the surface before points are lost, and weight (about 0.9 kg handheld versus about 4 kg for a structured-light head) determines whether the unit is comfortable for long handheld sessions. Optional RGB texture/color capture is only worth paying for when appearance matters.

Worked examples from published datasheets make the numbers concrete. The Creaform HandySCAN BLACK Elite states 0.025 mm accuracy, 0.020 mm + 0.040 mm/m volumetric accuracy, 0.025 mm resolution, 1,300,000 measurements/s, 22 blue laser lines (plus one), a 200 to 450 mm working distance, a 0.05 to 4 m part range, and 0.94 kg weight. The HandySCAN BLACK+ Elite pushes that to 1,800,000 measurements/s and 30 blue laser lines (plus one) over a 200 to 750 mm working distance, and with the Accu+ kit reaches 0.020 mm + 0.015 mm/m volumetric accuracy. The ZEISS/GOM ATOS Q is a blue-LED fringe-projection system whose twin 8- or 12-megapixel stereo cameras deliver up to 8 M or 12 M points per scan, point distance of 0.04 to 0.15 mm (8M) or 0.03 to 0.12 mm (12M), five measuring volumes from 100×70 to 500×370 mm, a working distance around 490 mm, a roughly 4 kg head, and Triple Scan for reflective surfaces. The Artec Eva, a structured-light unit that scans many parts without targets or powder, states roughly 0.1 mm 3D point accuracy, about 0.2 mm 3D resolution, up to 16 FPS and color texture — positioned for fast general-purpose capture rather than the tightest tolerance.

Selection Decision Factors

To turn the preceding five chapters into a specific model, follow the decision sequence below. Most selection mistakes come not from a single wrong step but from deciding too early at the wrong level — picking a brand before fixing the tolerance, or chasing resolution before settling the accuracy class. These nine factors can serve as a fixed RFQ template.

1. Tolerance rule of thumb: scanner accuracy should be roughly 3 to 10 times tighter than the tightest part tolerance you must verify (a gage R&R / 10:1 mindset). First-article inspection and metrology demand the tightest accuracy (about plus-or-minus 0.01 to 0.05 mm); machined-part reverse engineering typically needs plus-or-minus 0.03 to 0.10 mm.
2. Accuracy versus resolution: accuracy (closeness to true dimension) matters more than resolution for inspection and reverse engineering. Over-fine resolution makes large, noisy files that are harder to convert into clean NURBS surfaces.
3. Part size to volumetric accuracy: for large parts the "+ x mm/m" term dominates; pair scanning with photogrammetry and targets to bound accumulated error.
4. Surface and color: matte and cooperative surfaces favor structured light; dark and reflective parts favor blue-laser triangulation and Triple-Scan; if appearance matters, choose a unit with texture/color capture.
5. Technology fit: laser triangulation for industrial, reflective and high-accuracy handheld work; structured light for fast field capture on matte parts in controlled light; ToF or phase-shift for buildings and plants; laser tracker for large assemblies.
6. Stated versus accredited accuracy: prefer a VDI/VDE 2634 or ISO 10360-13 certificate from an ISO/IEC 17025 accredited lab, not just a marketing spec.
7. Workflow and software: CAD and metrology software compatibility (inspection deviation maps, GD&T, mesh-to-NURBS) and ease of recalibration.
8. Portability and connectivity: wireless versus tethered (the market is shifting toward wireless metrology), weight, battery life, and on-site versus lab use.
9. Environment: temperature stability, vibration, and ambient light — all of which degrade large-volume accuracy if uncontrolled.

One dimension worth stating explicitly is the governing standards a certificate should reference, because that is what separates a defensible accuracy claim from a marketing one. For optical area scanners the current framework is ISO 10360-13:2021, the acceptance and reverification test for optical 3D coordinate measuring systems, published in September 2021 and valid where surface gloss and color are within a cooperative range. VDI/VDE 2634 is the German guideline widely used for stated-accuracy certificates, with Part 2 covering single-view systems and Part 3 covering multiple-view self-referencing systems, defining probing error, sphere-spacing error and flatness-measurement error tests using calibrated spheres or artifacts. Laser trackers are verified to ISO 10360-10:2021 and ASME B89.4.19-2021, both revised in 2021 with a more sensitive interim test and a two-face health check, and NIST publishes interim-testing guidance against both. Underneath all of these, the calibration laboratory issuing the accuracy certificate must itself be accredited to ISO/IEC 17025:2017, which is what distinguishes stated from accredited accuracy; national variants and protocols include China's JJF 1951, and the broader ISO 10360 series (for example Part 1 terminology) provides the GPS/CMM context.

Finally, weigh manufacturer fit against the workload. Verified makers in this category include Creaform (AMETEK), whose HandySCAN and MetraSCAN lines are widely cited metrology-grade blue-laser systems, with the white-structured-light Go!SCAN serving fast color capture; ZEISS / GOM, whose ATOS (Q, LRX) blue-LED fringe-projection systems are regarded as a high-accuracy benchmark on small parts; Hexagon (Leica/Romer), with the Absolute Arm portable CMM and the Leica Absolute Tracker; FARO, with the Quantum / Quantum Max ScanArm, FARO Laser Tracker and Focus terrestrial scanners; Artec 3D, with the Eva, Leo and Space Spider handheld structured-light units; Nikon Metrology, with laser scanners, ModelMaker and large-volume metrology; and Shining3D, whose EinScan and FreeScan families span the entry-to-mid industrial tier (list pricing runs from roughly 2,000 USD for entry EinScan desktop units up to about 30,000 USD for metrology-grade FreeScan handhelds) and reflect the strong China supplier base relevant to this platform.