Bench record / observed, repeated, explained
The first useful scan is a controlled reference run. Its purpose is to expose installation errors while the scene, object, and processing choices remain simple. A decorative figurine may produce an attractive mesh, yet it cannot reveal whether scale survived export or whether a flat machined face was doubled during alignment.
A reference part gives the shop a project that can be repeated after moving the scanner, updating software, or changing the workstation. The value comes from the record surrounding the mesh as much as from the mesh itself.
Choose a part that can diagnose faults
Select a rigid object that fits comfortably inside the scanner’s documented field of view and working envelope. It should remain stationary on the turntable or fixture, with clearance around its full rotation. A medium-gray matte surface is a friendly starting point for many optical systems.
Useful geometry includes an asymmetric outline, a broad plane, a rounded feature, a through-hole, and a fine edge. Asymmetry helps feature-based alignment. EinScan’s fixed-scan guidance explains that parts with enough asymmetric geometric features can align efficiently, while repetitive patterns may fail the feature test.
Avoid transparent plastic, mirror finishes, deep glossy black, flexible parts, and repeated identical ribs for this run. Artec’s preparation guidance identifies black, transparent, and reflective surfaces as difficult cases and recommends powder or anti-glare treatment where suitable. Those materials can be tested later with a documented surface-preparation process.
Keep a short reason beside the chosen part: “rigid bracket; asymmetric; one bore; two measurable spans; untreated matte finish.” That reason helps a future operator replace it with an equivalent artifact if it is lost.
Record the physical reference
Clean the part without changing its surface. Mark the intended top and a repeatable mounting face with a removable tag placed outside the capture area. Photograph the part beside its project label.
Choose several dimensions that can be measured independently. Overall length and height are convenient for scale. A hole diameter tests a fitted feature. A step or slot width tests shorter spans. Record the exact measurement definition, contact points, instrument identifier, displayed resolution, date, and operator.
A shop caliper reading can support a commissioning comparison, but it does not make a traceability claim by itself. NIST’s discussion of dimensional traceability explains that repeated measurements of uncalibrated workpieces must be supplemented by calibrated artifacts when measurement bias and formal traceability matter. The first scan log should therefore call ordinary shop values “reference readings” unless the instrument and artifact chain support stronger language.
Prepare the scene with the fewest variables
Restore the bench, lighting, cable route, and scanner position documented during readiness checks. Confirm that the current calibration completed successfully. Use the direct tested data port and local project storage.
Mount the part so it cannot slide. Keep adhesive putty, clamps, or supports away from important surfaces. If the turntable uses coded targets, leave them visible; EinScan warns that an object covering those targets can prevent capture in target-based turntable alignment.
Choose the manufacturer’s recommended general or first-scan mode. Leave aggressive smoothing, automatic hole filling, and heavy decimation for a copy created after raw data is saved. Capture geometry first unless texture has a defined engineering use.
Set distance and exposure in that order
Place the part at the working distance indicated by the model’s software overlay or manual. Distance changes the camera geometry and field coverage; exposure cannot correct a part outside the usable range. EinScan’s fixed-mode workflow likewise establishes distance before brightness.
Open preview and rotate the empty or loaded turntable slowly through a full revolution. The part should remain within the useful view and should not hide all alignment features. Correct the mount before recording.
Adjust brightness or exposure until the software’s feedback shows usable surface data without large saturated or dark areas. Record the chosen mode and value exactly as displayed. A setting number has meaning only within that scanner, software version, and capture mode.
Capture one continuous baseline
Start recording with the most distinctive face visible. Watch the reconstructed model and tracking indicator rather than staring only at the physical part. Artec’s quick guide advises watching the object on screen and identifies fast scanner motion, small visible areas, and simple geometry as causes of lost tracking.
For a turntable scan, let the programmed sequence finish without touching the bench. For a handheld pass, move smoothly and maintain overlap. If tracking is lost, return gently to a recently captured distinctive region instead of sweeping forward.
Stop when the planned upper surfaces have solid coverage. Save this capture as the baseline raw project before changing the part. Record warnings, lost-tracking events, and any pause. A clean-looking preview should not erase a known interruption from the log.
Add the hidden side as a separate pass
Most mounted parts have a contact area the scanner cannot see. Photograph the first orientation, then turn the part onto the second documented face. Maintain overlap with the first pass by leaving distinctive side geometry visible.
Capture the second orientation as a separate scan. Artec’s documentation recommends scanning remaining regions after turning the object and preserving a previously scanned region to aid alignment. Keeping the passes separate also makes a bad underside capture easier to remove.
Align the passes using the software’s recommended method. Inspect the overlap before fusion. Parallel double edges, a step on a continuous plane, or two centers for one hole indicate misalignment. Return to the raw passes and alignment settings instead of smoothing the symptom.
Inspect the raw evidence before making a mesh
Rotate the aligned data under a neutral shading mode. Look for floating islands, streaks, clipped edges, large missing regions, and repeated surfaces. Compare every defect with the capture log.
- Missing data that rotates with a shiny face points toward surface response or exposure.
- A seam only in overlap points toward alignment or insufficient shared geometry.
- Regular wobble around the turntable suggests part movement or mechanical disturbance.
- A clipped side throughout the run suggests distance, framing, or working-envelope error.
- Random gaps accompanied by connection warnings return attention to the computer and USB path.
Change one condition for a diagnostic repeat. Label the repeat with that condition. The baseline remains untouched for comparison.
Create a conservative first mesh
Duplicate the project or processing branch. Apply the normal fusion or meshing preset with minimal smoothing. Remove isolated fragments that clearly belong to the fixture or background. Leave functional holes open and preserve fine edges for later inspection.
Automatic hole filling can make a model watertight while inventing a surface the cameras never observed. A filled bore or mounting face may look complete and still be unusable for reverse engineering. The first reference mesh should show where evidence ends.
Save the mesh with a name that identifies part, orientation set, software version, and processing revision. Keep the raw project, aligned data, and mesh together.
Finish with a repeat, not a beauty pass
Remove and remount the part using the recorded orientation. Run the same capture once more without changing exposure or processing. Compare coverage, alignment behavior, bounding dimensions, and the selected reference features.
Two similar results show that the installation can reproduce its own baseline. Two different results expose an uncontrolled variable: mounting, lighting, cable movement, tracking path, or processing. The next article’s dimensional workflow determines how large that difference is and whether the mesh is ready for a CAD handoff.