How to Control Accuracy When Scanning Large Parts: Markers, Photogrammetry, and Optical Tracking Compared

A 0.02 mm accuracy spec means something different on a 150 mm bracket than it does on a 6-meter wind turbine blade. On the bracket, that number describes the whole part. On the blade, it describes only a small patch of it, unless something ties every patch back to one shared coordinate system as the scanner moves down the length of the structure. That "something" is what separates the three main ways engineers control accuracy on large parts: physical markers, photogrammetry, and optical tracking.

Manufacturers publish this distinction as volumetric accuracy, usually written as a base value plus a per-meter term, such as 0.015 mm + 0.030 mm/m. In this published A + B mm/m expression, A is the constant term and B is the length-dependent term. Neither term should be relabeled as single-scan accuracy or treated as a complete uncertainty statement. On a small object the per-meter term barely matters. On a car body, a mold, or a ship hull, it can dominate the total error if nothing holds the coordinate frame steady over distance.
 

Why Local Accuracy Breaks Down Over Distance

A handheld 3D scanner without any external reference builds its position estimate from the part's own surface, matching each new frame of data to the frames just captured before it. That works well for a part small enough to keep the whole shape in view, because every new frame overlaps enough surface detail to register cleanly against the last one.

Stretch that same process across a part several meters long, and small registration errors from each frame start to stack. A fraction of a degree of drift in the calculated position, repeated over hundreds of frames, adds up to real deviation by the time the scanner reaches the far end. Flat, repetitive, or reflective surfaces make it worse, since there is less unique geometry for the scanner to lock onto frame to frame.

The fix is to give the scanner a stable, independent reference that spans the whole part, so each new frame registers against something fixed instead of against the last frame's own accumulated error. Markers, photogrammetry, and optical tracking are three different ways to build that reference.
 

Method 1: Physical Markers (Target-Based Registration)

The most direct method places reflective or coded targets on and around the part before scanning starts. A handheld laser scanner like SCANOLOGY's KSCAN-E registers against that target network the same way a surveyor works off fixed benchmarks, so the coordinate frame stays anchored to the markers instead of drifting with each new frame's own small errors.

On the KSCAN-E, the published volumetric accuracy is 0.015 mm + 0.030 mm/m in the standard configuration, tightening to 0.015 mm + 0.015 mm/m when an 800 mm scale bar is added to the marker layout. Because it's a target-based approach, it does not depend on any single external device tracking the scanner, only on the marker layout itself staying fixed relative to the part.

Markers are a proven method that needs no extra tracking hardware, and they work on surfaces that give a scanner little to grab onto otherwise, including flat panels and semi-gloss finishes. The tradeoff is time and surface access. Someone has to place enough markers to cover the whole part, then remove them afterward, and a part with a painted Class A surface or a cosmetic finish carries some risk of adhesive residue or witness marks from removal.

Method 2: Photogrammetry (Markers Plus a Camera Network)

Photogrammetry builds on the same marker concept, but adds a dedicated step to fix their exact positions before scanning even starts. Coded targets go across the entire structure, not just a local patch, and a photogrammetry camera photographs them from many angles around the part. Software triangulates those photos into precise 3D coordinates for every target, producing a coordinate network the scanner then registers against as it works.

SCANOLOGY's MSCAN-L15 photogrammetry system publishes a standalone volumetric accuracy of 0.012 mm/m and a shooting area of up to 9.4 m x 6.9 m, built for parts in the 2 to 10 meter range such as wind turbine blades, ship hulls, and aircraft wings. Paired with the KSCAN-E, the combined volumetric accuracy improves to 0.015 mm + 0.012 mm/m, reducing the published length-dependent term relative to the stated standard configuration.

A published SCANOLOGY case on a batch of 6 m x 1 m x 0.4 m wind turbine blade blanks shows the time cost and the accuracy payoff side by side. Reflective markers and coded points went on the blade first (8 minutes), followed by the MSCAN photogrammetry system capturing those points from multiple angles, then the scan itself (15 minutes) and report generation (5 minutes).

The scanner used in that case, HSCAN771, is an older SCANOLOGY handheld model since discontinued and no longer part of the current handheld lineup, and the photogrammetry hardware has since been succeeded by the current MSCAN-L15. Even with an older scanner, the combination increased accuracy by 67 percent and greatly reduced deviations in volumetric accuracy compared to running the handheld scanner alone on the same blade.

The added camera pass is what makes photogrammetry take longer to set up than markers alone. Photogrammetry can improve scale control relative to a target-only workflow on elongated or very large structures, where a marker layout without a camera network still leaves accumulated error uncorrected over the full length.

Method 3: Optical Tracking (Marker-Free)

Optical tracking flips what gets watched. Instead of the scanner locking onto targets fixed to the part, an external tracker locks onto the scanner itself, continuously reporting its position in real time as the operator moves it across the part. No targets go on the part's surface at all, and there's nothing to place or peel off afterward.

SCANOLOGY's NimbleTrack Gen2 delivers up to 0.025 mm system accuracy across both its C and E variants, entirely marker-free. The published specifications list 0.060 mm for NimbleTrack-C at 3.5 m, 0.059 mm for NimbleTrack-E at 3.5 m, and 0.072 mm for NimbleTrack-E at 4.2 m. Both offer an adaptive photogrammetry mode that holds volumetric accuracy at 0.044 mm + 0.012 mm/m for longer runs. The TrackScan Sharp series scales further, from the Sharp-49 variant at 6 m tracking distance and 49 m³ of coverage, up through the Sharp-S at 8.5 m and 135 m³, to the Sharp-Z at 12 m tracking distance and a 600 m³ measurement volume.

A SCANOLOGY case study on crane telescopic boom covers, ranging from 0.7 m to 14 m in length and weighing between 95 kg and nearly 4,000 kg, used the TrackScan Sharp-S in fully marker-free mode. Operating across its full 8.5 m tracking distance and 135 m³ measurement range, the product specifications list a maximum volumetric-accuracy value of 0.048 mm for the stated 10.4 m³ working volume, and completed a full inspection in about 10 minutes, with no marker placement or removal step in the workflow at all.

The gain in setup speed comes with one operational condition: the tracker needs an unobstructed line of sight to the scanner throughout the scan. On parts with deep pockets, internal cavities, or geometry that steps completely out of the tracker's field of view, the operator has to reposition the tracker partway through the job, which raises the question of how to keep everything in the same coordinate frame once that happens.

Comparing the Three Methods

Combining Methods for the Largest or Most Occluded Parts

These three approaches aren't mutually exclusive, and the largest or most geometrically complex parts often call for combining more than one. NimbleTrack Gen2's adaptive photogrammetry mode is a built-in hybrid, folding a lightweight photogrammetry pass into a marker-free tracking workflow to extend accuracy over longer runs. The TrackScan Sharp series offers the same kind of photogrammetry-assisted mode for tracking distances beyond roughly 6 meters, on top of its own marker-free tracking baseline.

Repositioning the tracker mid-job raises a different challenge: a second tracker position starts with its own independent coordinate frame unless something reconciles it with the first. DefinSight, SCANOLOGY's own scan-capture and metrology software, addresses this with a Super Coordinate Conversion (SCC) function that uses marker information together with shared point-cloud data to align measurements as the tracker moves from one position to the next, keeping the whole part in one coordinate system even when a single tracker position can't see all of it.
 

How to Choose for Your Part

Part size is the first thing to check. Markers alone typically cover smaller to mid-sized parts well; photogrammetry earns its extra setup time once a structure runs past a few meters in its longest dimension, which is why SCANOLOGY positions the MSCAN-L15 for the 2 to 10 meter range specifically. Optical tracking scales further still: the crane boom case above covered parts from 0.7 m to 14 m with the TrackScan Sharp-S, and the Sharp-Z extends tracking distance to 12 m across a 600 m³ measurement volume for the largest structures.

Surface finish matters next. A cosmetic or Class A surface where adhesive residue or witness marks are unacceptable points toward optical tracking, since nothing touches the part itself. A part that's going to be repainted or machined afterward removes that concern and opens up markers or photogrammetry as options.

Geometry and access decide the rest. A part with deep pockets, internal features, or shapes that block line of sight in multiple directions favors markers or photogrammetry, since both methods reference fixed points rather than requiring a continuous view of the scanner. A part that's large but open, like a vehicle body or a boom section, is a better fit for tracking, where the time saved by skipping marker placement adds up fast if the same part type comes through inspection repeatedly.
 

Frequently Asked Questions

What does a volumetric accuracy spec of "0.015 mm + 0.030 mm/m" actually mean?
The first value is the constant term and the second is the length-dependent term in the published expression. On a 3-meter part, that spec works out to roughly 0.015 mm plus three times 0.030 mm, not just the base 0.015 mm figure alone.

Do reflective markers damage or mark a cosmetic surface?
Standard photogrammetry and marker targets are designed to be removable without damaging most industrial finishes, but adhesive residue and light witness marks are a real risk on sensitive Class A surfaces, especially with longer scan sessions or warm environments. Parts where this is a concern are better suited to a marker-free optical tracking workflow.

Is photogrammetry still worth using if a job already has an optical tracker on hand?
Yes, for the largest runs. NimbleTrack Gen2 and the TrackScan Sharp series both offer built-in photogrammetry modes specifically because tracking accuracy alone starts to loosen at longer distances, and adding a photogrammetry pass tightens the per-meter error term back down for extended-length parts.

Can a single tracker position cover an entire large or complex part?
Not always. Deep pockets, internal cavities, or parts that wrap around out of the tracker's field of view require repositioning the tracker at least once. SCANOLOGY's DefinSight software handles that transition with its Super Coordinate Conversion function, which reconciles the new tracker position with the data already captured so the part stays in one coordinate system.
 

Getting the Right Setup for Your Part

The right accuracy-control method depends on the specific part, not a single headline spec. SCANOLOGY, the industrial 3D scanning brand of SCANTECH (Hangzhou) Co., Ltd., builds all three approaches into its product lineup, from the KSCAN-E and MSCAN-L15 for marker and photogrammetry-based workflows to the NimbleTrack Gen2 and TrackScan Sharp series for marker-free optical tracking, so the method can be chosen to fit the part rather than the other way around.