A 3D scanner's field of view (FOV) should be sized to the longest dimension of the part, not the bench or booth, and on sub-100 mm components this drives point density into the 5–20 micron range using structured-light desktop systems [S2][S3].
Small-part inspection covers roughly the 1–150 mm envelope: jewelry, electronic connectors, medical device hardware, injection-molded features, and small machined aerospace fittings all fall in this band, and each demands a different FOV-versus-accuracy trade-off than full-body or large-assembly scanning [S3][S5].
How FOV, point density and accuracy couple on small parts
FOV and point density are inversely coupled inside a fixed-resolution camera: a smaller FOV delivers higher points-per-mm², which is exactly what feature edges and tight tolerances need [S3]. On a desktop structured-light scanner, the smallest FOV option resolves features down to 5 microns and the larger FOV options relax to roughly 20 microns as the scan area grows [S2]. For a portable laser-line system sized to sub-meter parts, volumetric accuracy lands near 33 microns per meter with single-shot precision around 20 microns inside a 1 m envelope, which is the operating band most buyers compare against when the part is too large for a desktop rig [S2]. Two-camera geometry produces more reliable 3D measurements than single-camera geometry because the triangulation baseline resolves depth on low-texture surfaces, and this matters more as FOV shrinks [S1].
FOV classes mapped to typical small-part sizes
Industrial vendors publish FOV ranges tied to the optical bench, not to marketing labels. A common reference pair is the Artec-class small FOV at roughly 170–350 mm stand-off versus a larger 400–1000 mm FOV for medium parts [S1]. Desktop structured-light systems for very small parts drop into sub-100 mm scan volumes where 5–10 micron detail capture becomes realistic, and at the other end handheld laser-line systems scale to 1 m+ envelopes with multi-line laser crosses for full coverage [S1][S2]. The decision rule is mechanical: if the longest feature fits inside the smallest FOV that still keeps the part inside the scanner's stated Z-near to Z-far depth band, use it; if the part exceeds that FOV, step up one FOV class and accept the point-density penalty rather than tiling small-FOV scans [S1][S6].
Selection criteria for a small-part inspection scanner
Accuracy specification must be traceable to a known reference (calibration sphere, gauge block, or VDI/VDE 2634-style test) and reported with the FOV used, because vendors quote the best-case number on the smallest FOV [S3][S5]. Surface handling matters as much as FOV: shiny, dark and transparent parts defeat simple depth cameras and need a structured-light projector plus exposure bracketing, and blue-light projection is widely preferred in industrial metrology because its shorter wavelength supports sharper fringe contrast and tolerates more ambient light [S3][S5]. For a comparison of common options against decision criteria, the table below is the working spec frame engineers actually apply during procurement: structured-light desktop, structured-light portable, laser-line portable handheld, and blue-light industrial.
Comparison: scanner classes against small-part decision criteria
Four scanner classes cover virtually every small-part inspection request, and the selection hinges on four decision criteria: FOV range, stated accuracy, surface compatibility, and footprint [S1][S2][S3][S5]. Structured-light desktop rigs win on the smallest FOV and the best 5–20 micron accuracy band, but they need a controlled bench, vibration isolation, and a fixed stand-off [S2]. Portable structured-light adds a tripod-mounted, repositionable geometry that handles parts up to roughly 350 mm with multi-position registration, trading some accuracy for shop-floor flexibility [S6]. Handheld laser-line systems such as the FreeScan-class portable push coverage to 1 m+ with 20–33 micron volumetric accuracy and 26+ laser crosses, making them the default when the part is too big or oddly fixtured for a desktop setup [S2]. Blue-light industrial scanners (ATOS Q / ZEISS class) sit at the metrology end of the spectrum, delivering traceable accuracy, repeatable point clouds on freeform surfaces, and CAD-to-part comparison on castings, moulded parts and turbine hardware [S3][S5]. Across all four, the failure mode is the same: pushing a large-FOV scanner onto a sub-20 mm feature drops point density below what GD&T callouts require [S1][S3].
Workflow failures and corrective actions
Symptom: blurred edges or missing holes on a sub-10 mm feature. Root cause: the FOV is too large for the part, so point density across the feature falls under the resolution the GD&T callout demands. Corrective action: drop one FOV class or switch to a desktop structured-light system whose smallest FOV is documented at 5 microns; verify on a calibrated gauge before committing production scans [S1][S2]. Symptom: scan takes 3× longer than quoted and the mesh has registration artefacts. Root cause: a small-FOV scanner is being tiled across a part that should be on a mid-FOV class, multiplying merge steps and noise. Corrective action: re-class the part to the next FOV step, accept the lower per-point density, and add reference targets to hold volumetric accuracy [S1][S6]. Symptom: holes, gaps, or noisy clouds on shiny steel, dark anodized aluminium, or transparent polymer. Root cause: the projection wavelength and exposure are wrong for the surface reflectance. Corrective action: move to blue-light structured light with adjustable exposure and apply anti-reflection spray or fine chalk only as a last resort, because coating adds a systematic bias that has to be removed in post [S3][S5]. Symptom: per-scan accuracy passes the bench test but full-volume accuracy fails on a 1 m part. Root cause: the volumetric error budget is dominated by registration, not the single-shot specification. Corrective action: add photogrammetry scale bars, increase the count of coded reference targets, and re-measure against a traceable length standard before signing the inspection report [S2][S3]. When the symptom is a CAD-to-part deviation that drifts across the mesh rather than tracking features, replace the scanner rather than recalibrate: the optics have aged out of the original calibration volume and a re-cal will not recover the lost field flatness [S3].
Standards, sourcing and the inspection chain
Buyers should require the accuracy and FOV statement to be reported together, ideally referenced against a recognized metrology standard such as ISO 10360 or the VDI/VDE 2634 series, and the inspection software chain should be closed-loop with the CAD seed file to keep nominal-to-actual comparison auditable [S3][S5]. For a deeper look at non-contact distance measurement physics that overlaps with confocal and chromatic techniques, see the confocal vs chromatic comparison, and for a structured-light reference starting from the optical principles, the structured light scanner encyclopedia page covers projector and camera geometry. For shops already running a 3D scanner on the bench, an industrial barcode scanner on the same fixturing line lets part serials travel with the mesh, which closes the audit trail that small-part inspection reports are now expected to carry. Trackable signals for the next cycle: vendor release notes for sub-5 micron desktop FOV options in late 2026, and ISO 10360-style test reports published alongside the smallest-FOV accuracy claim rather than as a separate datasheet [S2][S3][S5].
Spec-level background on the components involved: 3d scanner.