Optical comparators, also called profile projectors, project a magnified silhouette of a workpiece onto a viewing screen or camera sensor and have remained a 2D, non-contact shop-floor measurement workhorse for roughly a century [S1]. A telecentric optical path preserves scale across the field of view, which is the property that lets a shop-floor operator measure threads, stampings, gears, and gaskets against overlays or CAD overlays without parallax error [S2][S5].
Selection in 2026 is driven less by the projector itself and more by the surrounding ecosystem: digital readouts, edge-detection software, LED illumination packages, and CNC stages that automate traverse between features. Buyers who ignore the configuration of the workpiece (flat vs cylindrical, light vs heavy, small vs large) typically overpay for capacity they never use [S1][S4].
Horizontal vs Vertical: Pick by Workpiece Geometry First
The single highest-leverage decision is horizontal versus vertical orientation, and it is set by how the part naturally sits, per Qualitest's 17 December 2025 guide [S4]. A vertical comparator illuminates from below through a glass stage, projects upward, and lets gravity fixture flat parts like gaskets, O-rings, stamped fabrications, and bare PCBs [S4]. A horizontal comparator mounts the optics to the side, projects laterally, and uses T-slot or dovetail stages with centers and clamps to hold shafts, threaded screws, and dense castings that cannot stand on their own [S4].
Workpiece envelope is the next gate: vertical machines are typically chosen when the largest dimension of the part fits comfortably on the stage glass, because loading is a place-and-measure action. Horizontal machines win when the part is long, heavy, or cylindrical, because the side-mounted lens path keeps the silhouette in focus along the entire axis without the operator having to flip the part [S4]. For buyers evaluating both, the optical comparator reference page lays out the underlying terminology (telecentricity, field of view, screen diameter) used in every OEM datasheet.
Magnification, Screen Size, and Field-of-View Math
Magnification is the headline spec, but field of view is what determines whether a part actually fits on the screen, as IQS Directory notes in its 31 August 2026 update [S2]. The visible area is set by dividing the screen diameter by the magnification: a 300 mm screen at 10x shows a 30 mm field of view, while the same screen at 50x shows only 6 mm, which is fine for a thread flank but useless for a stamped bracket [S2]. Operators usually spec a 10x lens as the workhorse and add a 20x, 50x, or 100x lens for fine features, then confirm the chosen lens covers the largest dimension of the typical part.
Screen diameter ties directly to the largest feature you need to inspect in one shot. 300 mm (12 in) and 400 mm (16 in) are the common shop-floor sizes; 600 mm and larger are reserved for long shafts and large stampings where the part must stay in one silhouette to preserve measurement traceability [S2][S4]. For buyers cross-shopping alternatives such as a vision measuring machine, the field-of-view math is the same constraint, so the comparator remains attractive only when the part is genuinely 2D and fits the screen.
Digital Readout, Edge Detection, and Throughput
A modern optical comparator is no longer just a projector and an overlay; the digital readout (DRO), automatic edge detection, and CNC stage are what separate a 1990s bench unit from a 2026 shop-floor cell. Q-Check DROs and Projectron automatic edge-detection overlays convert a manual silhouette trace into a parts-per-million-grade measurement, and LED ring lights (such as OGP's TruLight) replace hot halogen bulbs to stabilise edge contrast across shifts [S1]. For high-mix shops, the software layer (ZONE3, Measure-X, or EVOLVE on the OGP side) is the real productivity multiplier because it lets an operator run an inspection routine without re-touching the stage between parts [S1].
Throughput in production is dominated by the time to load, focus, and traverse between features, not by the projection step itself. CNC stages with motorised X-Y and rotary axes cut that traverse time, and digital edge detection removes the operator-to-operator reading variance that historically made comparator data noisy on second shift [S1][S2]. For buyers weighing a profile projector against a digital panel meter for a metrology bench, the comparator's role is dimensional inspection of a physical part, not electrical readout, so the comparison is a category error.
Calibration, Standards, and What "Accurate" Actually Means
An optical comparator's accuracy is bounded by lens calibration, stage straightness, and screen or scale linearity, and is normally traceable to a national metrology institute through the manufacturer's calibration certificate [S2]. Buyers should request the calibration uncertainty at the working magnification, not the headline "micron accuracy" figure, because the same instrument can read 5 micrometres at 10x and 15 micrometres at 50x on a single screen [S2][S5].
Two practical rules from the field. First, an optical comparator avoids parallax error because the optics are telecentric, but it does not avoid cosine error when a feature is not parallel to the stage; the operator must fixture the part so the dimension of interest lies flat [S5]. Second, a comparator is a 2D instrument: it will not certify a 3D GD&T callout, and a CMM or multisensor system is the right tool when the part has true 3D tolerance stacks [S1]. Where optical-path components matter, the underlying optical glass grades used in the lens stack drive chromatic correction, which is why low-end comparators fringe at high magnification.
Selection Criteria Mapped to Part Type
Four realistic configurations cover the bulk of 2026 shop-floor demand, and the choice is driven by three criteria: part orientation, weight, and required magnification. A buyer who locks these three numbers first rarely gets the spec wrong. [S5]
Configuration A, manual vertical with 300 mm screen and 10x fixed lens, fits flat gaskets, O-rings, and small stampings under about 100 g where place-and-measure is acceptable [S4]. Configuration B, digital vertical with CNC X-Y stage and Projectron-style edge detection, fits PCB panels and high-volume stampings where throughput and operator-to-operator repeatability dominate [S1][S4]. Configuration C, manual horizontal with T-slot stage, fits shafts and threaded fasteners under about 20 kg where the part must be held between centres [S4]. Configuration D, CNC horizontal with rotary indexer and ZONE3-style software, fits turned components and small gears where angular features must be measured across multiple orientations without re-fixturing [S1][S2].
For shops that already own one of these and are adding capacity, the cheapest upgrade is usually a digital edge-detection retrofit and an LED ring light, not a new machine. For shops buying their first comparator, the expensive mistake is specifying a 600 mm screen and 50x magnification to handle a workload that 95% of the time lives inside a 300 mm screen at 10x, so the spec should follow the median part, not the worst case [S2].
Limitations and When Not to Buy a Comparator
Hexagon's resource page is blunt on the limits: an optical comparator measures a shadow, so any feature that is hidden from the projection direction is invisible to the instrument, and 3D features such as draft angles, true-position stacks across non-parallel planes, and wall thicknesses on enclosed features require a CMM or CT scanner instead [S3]. A comparator is also a single-magnification-per-lens tool on most models, so a part that needs 10x for the body and 100x for a micro-feature forces a lens swap that breaks measurement traceability between steps [S2][S5].
Two more failure modes from the field. First, comparator accuracy degrades as the part leaves the calibrated focal plane, so tall stampings and bowed sheet metal need a horizontal rig or a refocusable vertical stage; forcing them onto a fixed vertical glass distorts the silhouette. Second, comparator lighting is the most common silent error source: a halogen bulb that drops 20% in output across a shift moves the apparent edge, which is why 2026 shop-floor units ship with LED illumination specified to a colour temperature and a lifetime in hours, not just "brightness" [S1][S2].
Sourcing Logic and a Shortlist Approach
Three steps produce a defensible shortlist. First, list the ten most-produced parts and record the orientation, weight, and the largest single dimension that must fit one screen at the median magnification; this single worksheet removes roughly half the catalog from consideration. Second, separate the workload by horizontal-versus-vertical bias, because very few shops genuinely need both, and the one that does usually buys a horizontal unit first for shaft work and a smaller digital vertical unit for the bench. Third, validate the digital-readout and edge-detection package against the operator skill mix: a manual unit with an overlay still makes sense for a one-operator job shop, while a CNC unit pays back only when inspection routines are repeated across shifts [S1][S4].
For 2026 sourcing cycles, the practical signal to watch is the rollout of LED illumination retrofits and automatic edge-detection overlays onto existing comparators, because that path extends the life of installed units and defers a five-figure capital purchase. The other signal is the convergence of vision-measuring-machine software onto traditional profile-projector hardware, which is closing the throughput gap between a comparator and a small VMM for many 2D jobs [S1][S2].
Spec-level background on the components involved: pressure transmitter.