Slide 003: Point Clouds, Accuracy Specifications, and Common Artifacts

Slide Visual

Slide Overview

This slide covers the data output of 3D scanners — point clouds — and teaches students how to interpret scanner specifications and diagnose common scanning artifacts. Understanding these concepts is essential before moving into scanner operation (Module 2) and mesh processing (Module 3).

Instruction Notes

Point Cloud Data Structure

A point cloud is the fundamental output of any 3D scanning process. Each point contains:

• Position: X, Y, Z coordinates in the scanner's coordinate system (typically in millimeters)
• Color (optional): RGB values captured by the scanner's camera, enabling textured reconstruction
• Normal vector (optional): A unit vector indicating the estimated surface orientation at that point, critical for mesh reconstruction algorithms
• Intensity (optional): The strength of the reflected signal, useful for surface classification

Point clouds are stored in standard formats: PLY (Polygon File Format), PTS, XYZ, E57, and LAS. File sizes scale linearly with point count — a 1-million-point scan with XYZ+RGB data is approximately 20-30 MB in ASCII format, far less in binary.

Scanner Specification Interpretation

When evaluating a scanner, the following specifications determine fitness for purpose:

• Accuracy: How close measurements are to true values (e.g., ±0.05 mm). Measured against calibrated reference objects per VDI/VDE 2634 or ISO 10360 standards.
• Resolution/Point spacing: The minimum distance between adjacent points (e.g., 0.1 mm). Determines the smallest capturable feature.
• Repeatability: Variation between repeated scans of the same object (e.g., ±0.02 mm). Lower is better.
• Working volume: The 3D space the scanner can capture in one setup (e.g., 200 x 150 x 150 mm). Objects larger than the working volume require multiple scans and alignment.
• Scan speed: Points per second or frames per second. Affects workflow efficiency.

Common Scanning Artifacts

Students must learn to identify and diagnose these artifacts:

• Noise: Random scatter of points around the true surface, caused by sensor limitations, vibration, or poor lighting. Visible as a "fuzzy" point cloud.
• Occlusion holes: Missing data where the scanner could not see the surface — deep cavities, undercuts, behind features.
• Registration errors: Misalignment between multiple scans, creating double-wall or stepped artifacts at scan boundaries.
• Reflectance artifacts: Bright spots or missing data on shiny surfaces where specular reflection overwhelmed the sensor.
• Edge noise: Exaggerated scatter at sharp edges where the scanner struggles to resolve the transition between surfaces.
• Motion blur: Smeared geometry caused by object or scanner movement during capture.

Artifact Mitigation Strategies

Artifact	Cause	Mitigation
Noise	Sensor limits, ambient light	Average multiple scans, increase exposure
Occlusion	Line-of-sight blockage	Multi-angle scanning, repositioning
Reflectance	Shiny surfaces	Dulling spray, scanning powder, polarization
Edge noise	Sharp transitions	Higher resolution settings, closer range
Motion blur	Movement during scan	Secure fixturing, vibration isolation

Key Talking Points

• Point clouds are raw data — they must be processed into meshes for most applications
• Scanner accuracy and resolution are independent specifications — both matter
• Every scan will have artifacts — the skill is in diagnosing and mitigating them
• VDI/VDE 2634 is the standard for evaluating scanner accuracy claims
• File format choice affects downstream compatibility and file size

Learning Objectives (Concept Check)

• Define the components of a point cloud data point (XYZ, color, normal, intensity)
• Interpret a scanner specification sheet and explain accuracy vs. resolution
• Identify at least four common scanning artifacts from visual examples
• Recommend mitigation strategies for noise, occlusion, and reflectance artifacts

---

Last Updated: 2026-03-19