The terms optical comparator, profile projector, and shadowgraph describe the same non-contact 2D metrology instrument across at least four independent reference sources [S1][S3][S4][S5], and there is no functional difference between the three names, only in regional usage and in the features bolted onto the bench [S5].
A profile projector shines light past a workpiece, magnifies the silhouette through a telecentric lens, and projects the shadow onto a ground-glass screen so an operator can measure it against a reticle, scale, or mylar overlay [S1][S2]. The dedicated shadowgraph-only projector is simply the same optical chain sold or configured without the optional measurement stage, edge sensor, or CNC axes, so its job stops at visual comparison against a chart.
Light path and silhouette: identical physics, both names
The optical path is light source, condenser, workpiece on a glass stage, telecentric projection lens, mirror, and translucent screen, with telecentric optics critical because they project parallel rays and keep magnification constant as the part shifts in Z [S2][S5]. A general lens would let the operator judge perspective, but a telecentric lens projects the same size for near and far objects, which is what stops a 0.5 mm feature from ballooning into a different number when the operator refocuses [S2].
Two illumination modes are standard on either configuration. Contour illumination, transmitted from below, produces a crisp dark silhouette on a bright background and is the mode used for thread form, edge location, gear teeth, and external profile [S2][S5]. Surface illumination, reflected from above, projects the upper surface and is used for inspecting print, scribed lines, recessed features, and the inside of holes or counterbores [S2]. The shadowgraph-only projector is normally run in contour mode because that is what the name implies, while a full optical comparator carries both modes plus blackout curtains for shop-floor ambient light [S2].
Magnification, screen, and stage travel: where the two diverge
Magnification lenses typically range from 10x to 100x, with a 20x lens turning a 0.5 mm feature into a 10 mm shadow on the screen [S1], and bench-top units commonly ship with 10x, 20x, 50x, and 100x sets that an operator swaps via a turret or lens carrier [S1][S5]. Large floor-standing comparators carry screen diameters that exceed 1 m, which is why the format survives for very large stamped or turned parts [S2].
The split shows up in the stage and readout. A dedicated shadowgraph-only projector carries a basic glass stage and a manual chart overlay, with the operator reading a length by eye against a scale or a printed mylar of the part drawing [S1][S5]. An optical comparator adds a precision X-Y stage driven by micrometer heads or motorised axes, a digital counter reading X and Y to micron resolution, and a screen that rotates with a protractor scale for angle measurement [S5]. The shadowgraph-only configuration is the cheaper path for high-mix go/no-go or visual comparison, while the full comparator is the path when the operator needs a numeric readout for SPC and a paper trail.
Measurement methods: overlay, stage DRO, and optional edge detection
Three measurement methods are used in practice. The first is a graduated screen with a reticle laid over the silhouette, equivalent to a clear ruler placed on the image [S3]. The second is the stage method, where the operator lines up the screen cross-hair with one edge, zeros the digital readout, drives the stage until the cross-hair meets the other edge, and the travel of the stage is the real dimension because the stage moves the part, not the magnified image [S1][S3]. The third is software-based image analysis that finds edges and reports measurements, which is newer and not as widespread as the first two but is being adopted as digital readouts and cameras get cheaper [S3].
A shadowgraph-only projector normally runs method one and method two on a manual stage, with a print of the part outline taped to the screen for comparison [S1][S5]. A full optical comparator layers on method three through an edge-detection sensor or vision system, which lets the unit measure 50 points on a gear tooth in the time an operator would spend on five. When properly calibrated, an optical comparator system can hit positioning accuracies around 0.005 mm against a calibrated overlay or stage scale [S4], and accuracy does not depend on the operator judging distance by eye because the reading comes from stage travel or the overlay, not from the magnified image [S1].
Vertical vs horizontal: same optics, different gravity axis
Two main configurations exist for both formats. Vertical comparators, where light travels upward through a horizontal stage, are preferred for flat or small parts that can be easily placed on the stage, including rubber gaskets, plastic mouldings, thin stampings, and small electronics [S4][S6]. Horizontal comparators, where light travels horizontally past a part held between centres or on a V-block, are specified for heavy, tube-shaped, or long parts that would sag or be hard to fixture on a flat stage [S4][S6].
Fixture strategy follows the configuration. A round shaft goes between centres or onto V-blocks on a horizontal unit so its centreline sits on the optical axis, while a flat stamped bracket gets clamped to the vertical stage with clips, clamps, or magnets so it does not shift mid-measurement [S2][S6]. A shadowgraph-only projector is most often seen as a vertical bench unit for small parts, while a horizontal comparator with a long throw and a large rotary screen is the format of choice for shafts, threaded fasteners, and long cutting tools where the whole length has to be in focus at once.
What the two actually own: 2D features, threads, gears, stampings
Optical comparators are widely used for complex-shape stampings, gears, cams, threads, and comparing the measured contour to a model, and they sit across precision machinery, aerospace, watches and clocks, electronics, and instrumentation [S3]. The non-contact rule is exactly why these machines win for soft or delicate work: a bouncy silicone O-ring measured with metal calipers gets squashed, but a light beam leaves the part round and untouched [S4].
Typical use cases line up cleanly. A shop inspects a small 2D feature such as a thread profile, gear tooth, stamped contour, molded radius, or the tip of a cutting tool, then reads it on a calibrated overlay [S1][S3]. Multiple operators can stand around the screen at once, which is something a measuring microscope with an ocular cannot do [S2]. A dedicated shadowgraph-only projector handles the same outlines but skips the CNC stage, the rotary screen, and the edge-detection software, which is why it is the cheaper option for a cell that only does visual comparison against a chart.
Limits, standards, and what to do instead
The known limit is the 2D silhouette itself. The instrument was originally developed to inspect outlines, and models with measurement functions appeared later, so anything that depends on a true 3D surface, a hidden internal feature, or a depth dimension belongs on a coordinate measuring machine or a vision measuring system, not on a comparator [S2][S5]. For the most demanding sub-0.005 inch features, the practical limit comes back to magnification and screen resolution, which is covered in detail for sub-0.005 inch feature work.
The accuracy framework that profile projectors are specified against is JIS B 7184, which is the Japanese industrial standard for optical comparators and the reference most international datasheets cite for magnification accuracy, screen flatness, and stage linearity [S5]. A comparator will sit next to a coordinate measuring machine and a caliper and micrometer set on a working bench, but the three tools are not interchangeable, and the comparator is the right call only when the job is a clean 2D outline that can be backlit. The single next node to track is whether the inspection plan is really a 2D profile check: if yes, a shadowgraph-only projector will cover the comparison and a full optical comparator will cover the readout, and the two should not be specified as if they were different categories of instrument.