← RETURN TO INSIGHTS

The Pitfalls of Point Density in Industrial CT Scanning

Why more points don't always mean better data. We explore the physics behind voxel resolution, beam hardening, and how to filter point clouds accurately for reverse engineering.

When clients receive a point cloud containing 50 million data points from a high-resolution X-Ray Computed Tomography (CT) scan, the instinct is to believe they possess a perfect digital twin. This is a dangerous assumption that often leads to catastrophic downstream engineering errors.

Voxel Size vs. Structural Resolution

The root of the misunderstanding lies in confusing voxel size with structural resolution (the ability of the system to distinguish between two closely spaced features). Just because you can output a point cloud with a dense grid spacing doesn't mean the points accurately represent the boundary of the physical part.

Surface Determination Uncertainty
Figure 1: The uncertainty of surface boundaries within the voxel grid space.

The Beam Hardening Artifact

When scanning dense materials (like titanium or steel tooling), lower-energy X-ray photons are absorbed at the edges of the part, while higher-energy photons pass through to the center. This causes the edges of the recreated 3D volume to appear artificially denser than the core—a phenomenon known as beam hardening.

If left uncorrected by hardware filters (such as copper or tin plates at the x-ray tube) or software algorithms, the surface determination algorithm in software like VGStudio MAX will incorrectly interpret this density gradient. The resulting 3D mesh will show "cupping," where flat surfaces appear concave.

"Attempting to fit a theoretical GD&T cylinder to a CT scan exhibiting beam hardening will inevitably result in a rejected part, not because the part is bad, but because the scan data is warped."

Intelligent Filtering is Mandatory

To utilize CT data for true metrology grading, the point cloud must be intelligently filtered. Throwing millions of points at a CAD comparison software without decimation and noise-reduction leads to extreme processing delays and false non-conformities caused by scatter radiation noise being interpreted as surface deviations.

Our engineers utilize iterative filtering algorithms that preserve sharp geometric edges while smoothing high-frequency noise from planar surfaces, yielding an optimized mesh ready for rigorous ASME Y14.5 profile evaluation.

Trust your non-destructive data.

View NIKON CT Capabilities