Can 3D Scanners Accurately Capture Deep Holes? When to Add Probing or CAD Reconstruction

Optical 3D scanners can capture a deep hole accurately, but only up to a point. Handheld scanners built with a dedicated deep-hole scanning mode can image bores, grooves, and recesses that would otherwise sit in shadow, and for most inspection and reverse engineering work that's enough. The point where scanning alone stops being enough is a bore that's deep and narrow enough, especially one that's fully closed off at the far end, that light simply can't bounce back out of it to a camera. Past that point, the accurate answer isn't "buy a better scanner," it's "add a contact probe, or let CAD reconstruction fill in the geometry the scan can already infer."

This is a question that comes up constantly in quality inspection and reverse engineering: an engine cylinder bore, a mold's ejector pin hole, a bearing bore in a cast housing, a deep counterbore on an aerospace bracket. Whether scanning alone gets you a usable result, or whether you need to bring in probing or CAD reconstruction, comes down to the physics of how optical scanners actually see, and how deep the feature is relative to how wide it is.

Why Deep Holes Are the Blind Spot of Optical Scanning

Handheld 3D scanners, whether they use blue laser lines or structured light, work by triangulation. A projector throws a pattern (lines or light) onto the part, and one or more cameras, offset at a fixed angle from the projector, watch how that pattern distorts across the surface. The software calculates each 3D coordinate from the triangle formed by the projector, the camera, and that surface point. That principle only works if the same point on the surface is visible to both the projector and the camera at the same time.

Inside a hole, that stops being true. As the surface curves away into the bore, the projected light either can't reach the inner wall at a workable angle, or it does reach the wall but the reflected light never makes it back to the camera because the hole's edge blocks the line of sight. The deeper the hole relative to its diameter, the more of the interior falls into this shadow zone. A shallow, wide counterbore is easy; a long, narrow bore is hard, for the same reason a flashlight beam can't fully light the inside of a narrow tube from outside it. This is a well-documented limitation of triangulation-based scanning generally, not something specific to any one brand of scanner.

Threads, undercuts, and a blind bottom with no exit on the far side all compound the problem: even the small amount of light that does bounce off the inner wall may never find its way back out at an angle the camera can register.

What a Deep-Hole Scanning Mode Actually Changes

Recognizing this limitation, scanner manufacturers have built dedicated modes to push that shadow line further down the bore. SCANOLOGY's KSCAN-E includes a deep-hole scanning mode that switches from its standard multi-line laser pattern to a single blue laser line, built specifically to reach "hard-to-measure areas such as engine cylinders and internal grooves of cast parts," per the product's own documentation. Dropping to one active line isn't a downgrade for this purpose. It deliberately trades raw scan speed for better access to geometry a full multi-line pattern can't reach.

SCANOLOGY's SIMSCAN-S Gen2 addresses the same problem in two ways: it carries the same single-extra-line deep-hole scanning mode as KSCAN-E, and it's also built around a short-baseline camera design that, per the product page, lets it "capture gaps, deep holes, slots, and channels with a steep viewing angle" that a wider-baseline system would miss. At 0.015 mm accuracy and 560 g, its compact, lightweight head is also easier to angle down into the space above a bore by hand, which matters just as much as the optics when the real obstacle is physical clearance rather than resolution.

SIMSCAN-S Gen2 scanning a vehicle door panel, including a mounting hole, using its short-baseline camera design built for steep viewing angles.

This is why "deep hole" isn't a single category engineers should treat the same way every time. A wide, shallow counterbore is straightforward for either scanner's standard mode. A cylinder bore or an internal groove, where the depth is significant relative to the diameter, is exactly the case these dedicated modes exist for. Somewhere past that, geometry wins regardless of scanner design: for any external handheld triangulation scanner, an especially deep and narrow bore, particularly one that's closed off at the far end, will still have a floor and lower-wall region no line-of-sight optical method can reach. That's the point where the practical options become contact probing or, for fully enclosed internal geometry a probe can't reach either, industrial CT scanning.

When Scanning Alone Isn't Enough: Bring in a Probe

For holes that genuinely sit outside what optical scanning can resolve (a deep or fully closed bore, or a bore buried inside an assembly where the scanner physically can't get a clear sightline), contact probing solves a different problem than scanning does. A touch probe doesn't need to see the surface from an angle; it only needs to physically reach it and record where the tip made contact. Line of sight stops being the constraint. Reach does.

SCANOLOGY's AccuArm is built to physically reach and sample those surfaces. It's a portable, articulated-arm coordinate measuring machine (a PCMM), not a fixed CMM, and the distinction matters: a fixed CMM, whether it sits in a metrology lab or is installed directly in a production cell, is a stationary machine built around bringing the part to it, while a PCMM like AccuArm is carried to the part, whether that's a housing on an assembly line or equipment too large to move. AccuArm's 360-degree joint rotation is specifically what lets an operator angle the probe tip down into a bore or recess that a rigid measurement path couldn't reach, and its swap-probe design lets you switch probe styles without recalibrating, useful when a job mixes surface points with a deep-bore diameter check in the same session.

AccuArm is certified to ISO 10360-12, with published performance data broken out across several metrics, including single-point articulation performance (SPAT), probing size error, and probing form error. For the 6-axis Superior-grade configuration, SPAT specifically is 0.012 mm at 1.5 m reach, scaling to 0.055 mm at the arm's maximum 4.5 m reach, per SCANOLOGY's specification data. Those numbers describe the arm's own metrology performance under ISO test conditions rather than a guaranteed uncertainty for any specific bore (real-world results also depend on probe calibration, sampling strategy, and part condition), but they give a concrete, standards-based reference point that a scanner's optical accuracy spec alone doesn't provide once you're taking discrete contact points instead of a surface capture.

AccuArm's touch probe recording a contact point on a fixture, the kind of measurement that reaches past line-of-sight limits.

In practice, this doesn't mean choosing probing instead of scanning for the whole part. The common workflow is scanning everything the optics can reach, since that's dramatically faster for full-surface coverage, and reserving the probe for the specific bores or features the scan flagged as incomplete. SCANOLOGY positions AccuArm and its handheld scanners to work together as "a complete, unified measurement solution on DefinSight," per the AccuArm product page, rather than as two separate tools producing two separate reports an inspector has to reconcile by hand.

When the Answer Is CAD Reconstruction, Not More Data Points

Probing is the right call when you need to physically confirm a real-world measurement: a bore's actual diameter, its position relative to a datum, or its tolerance status. Reverse engineering has a different goal, producing a clean, editable CAD model of the part, and for that goal, chasing every last data point inside a bore isn't always the right use of time.

If a bore is a simple, regular feature, a straight cylindrical bore or a circular counterbore, a scan that captures the rim and enough of the upper wall to establish a reasonable diameter and axis often gives a CAD engineer enough to rebuild the feature as clean, parametric geometry, rather than trying to mesh every point of a shadowed lower wall. How much confidence that fit deserves depends on how much of the wall is actually visible and how clean that data is; a thin sliver of coverage can still produce an unreliable fit, which is exactly why a feature that needs a verified diameter, not just a modeled one, is a probing job, not a CAD-fitting job. SCANOLOGY's DefinSight MODEL is built for the modeling side of this: aligning scan data, sketching directly on the mesh, and fitting surfaces to rebuild the part as "fully parametric hybrid" CAD geometry rather than raw triangles. For a regular bore with enough visible wall data, that typically means sketching the circle the rim and upper-wall geometry define and building the feature parametrically from there, the same technique any experienced reverse engineer uses regardless of software, rather than trying to capture the literal, possibly incomplete, mesh data all the way to the bottom. What that approach can't do is recover the actual depth, termination, or any hidden step near the bottom of a bore the scan never saw; those need to come from a print reference, additional measurement, or probing, not from extrapolating the visible geometry.

The DefinSight MODEL workflow: raw scan data is aligned, surface-fit, and rebuilt into parametric CAD, shown here on a knuckle part with several bored features.

One real limit is worth being direct about: good reverse engineering practice means only fitting geometry the scan has actual evidence for. That discipline can extend a clean cylinder from partial wall data, but it draws the line at inventing a feature the scan never captured any trace of: an internal step, an undercut, a defect deep in the bore that no rim or partial-wall data hints at. That line comes from the judgment of the engineer doing the reconstruction; the software itself will let you sketch and extend geometry however you tell it to. If a hole is irregular, or if the reverse engineering goal is to catch an actual internal defect or anomaly rather than rebuild a nominal, regular feature, that calls for probing known locations, or, more reliably for an unknown or fully enclosed defect, industrial CT scanning. Assuming a surface fit will find it is the wrong bet. CAD reconstruction earns its keep completing a regular feature from partial, visible data. It's never a substitute for measuring a feature you actually need to verify, or one with genuine unknowns hidden inside it.

Matching the Method to the Job

A Combined Workflow, in Practice

A cast housing with several bearing bores is a good example of how these methods stack rather than compete. Scanning the housing's exterior and open faces with a handheld scanner in deep-hole mode captures the bulk of the part quickly and gets you well down into each bore's upper wall. For the bores that need a verified position and diameter check against print tolerance, AccuArm's probe confirms those specific features with contact measurement, paired with the scan data through DefinSight as part of SCANOLOGY's unified measurement workflow, rather than as two disconnected datasets. If the end goal is a reverse-engineered CAD model of the housing rather than an inspection report, an engineer working in DefinSight MODEL can use that combined scan-and-probe data to rebuild the bores as parametric cylinders, faster and cleaner than trying to mesh a feature that was only ever partially visible to the scanner in the first place.

None of the three steps replaces the other two. Scanning covers the part fast; probing reaches what scanning can't see; CAD reconstruction turns partial data into a usable, editable model. Deciding which combination a given part needs comes down to how deep the feature actually is, what you need to confirm about it, and whether the end deliverable is a measurement report or a CAD file.

FAQ

Can a 3D scanner measure a hole's diameter at all, or does it always need a probe? For holes with a clear opening and a depth-to-diameter ratio the scanner's deep-hole mode can handle, yes, a scan alone typically produces a usable diameter and position result. How closely that result tracks the scanner's headline accuracy spec also depends on how much of the wall is actually visible, alignment, and fitting method, not the spec number by itself. A probe becomes the better call once the feature is deep, narrow, or closed off enough that the optical method can't see far enough down the wall to fit a reliable diameter.

Is AccuArm a replacement for a fixed CMM? No. AccuArm is a portable CMM (PCMM): a lightweight articulated arm you carry to the part, wherever it sits. A fixed CMM, whether it's in a climate-controlled metrology lab or installed directly in a production cell, is a stationary machine built around bringing the part to it instead. The two aren't interchangeable categories; many shops use both: a fixed CMM for lab-based or high-volume final inspection, and a PCMM like AccuArm for in-process checks and hard-to-reach features where moving the part isn't practical.

Does CAD reconstruction invent geometry the scanner never captured? Done correctly, it shouldn't. Tools like DefinSight MODEL give an engineer the sketching and surface-fitting tools to fit clean, parametric geometry to the data a scan actually captured, so a cylindrical bore with a visible rim and partial wall becomes a clean cylinder feature: fitting a known geometric shape to real evidence, rather than guessing at geometry with no data behind it. That said, the software itself won't stop you from sketching further than the data supports; keeping the model honest is the engineer's call. Anything the scan genuinely didn't capture, an undercut, an internal step, an unexpected defect, needs to be measured directly by probe or CT, not assumed from the surrounding surface.

What's a practical sign that a hole needs probing instead of just a deeper scan mode? If repeated scan attempts, even in deep-hole mode, keep leaving a gap or noisy data at the bottom of the bore, the limitation is geometric, not a scanning technique problem, and it's time to bring in a probe.

Deep holes are one of the clearest cases in industrial metrology where the right answer isn't "buy the highest-accuracy scanner," it's matching the measurement method to what the geometry actually allows. A dedicated deep-hole scanning mode on a handheld scanner like KSCAN-E or SIMSCAN-S Gen2 gets you further into a bore than a standard scan pattern will. When a hole is deep, narrow, or blind enough to defeat that, a portable CMM like AccuArm reaches where optics can't, and when the goal is a CAD model rather than a measurement report, software like DefinSight MODEL turns partial scan data into a clean, usable feature. The right call comes down to how much of the bore is actually visible, what you need to verify about it, and whether you're handing off a measurement report or a CAD file.