1. Introduction
  2. Comparison
  3. Identification
  4. Translation
  5. Reflection


Identification


To interpret the scanning anomalies observed at Stratford Westfield, I developed five sets of indication.（Each category is defined for the purpose of this project analysis. They are not intended to be universal definitions.）




Data Gaps

Missing polygon errors in 3D scans present as voids where no surface data is recorded. In the Stratford–Westfield scan, this is most obvious on glass surfaces—since the scanner fails to detect glass, large gaps appear in the mesh. Such gaps occur whenever light passes through or around a material without returning to the sensor, a phenomenon common with transparent or highly absorptive materials. As a result, certain architectural elements (windows, open doors, bodies of water) become “invisible” in the digital model.






Surface Distortions

In the 3D scan of Westfield’s curved glass surfaces, straight lines are sometimes captured as wavy, flowing forms. These “melting architecture” artifacts appear when the scanning system receives conflicting or noisy data—for instance, multiple reflections from a curved glass canopy—and then attempts to average them. The algorithm struggles to reconcile these signals, producing approximations that look liquefied or warped. The scanner’s viewpoint can break a single straight line into irregular geometries and fold reflective surfaces into unrecognizable shapes.






Misregistered Geometry

When the same physical element is scanned multiple times from slightly different positions, duplicate-surface errors occur, creating “ghost” geometries that overlap the original. In a 3D mesh, this looks like a faint double layer on a wall or column. At Westfield, this commonly happens on glass surfaces that were captured from different angles.




Motion Artifacts

Whenever moving people or objects are captured during scanning, motion artifacts appear, resulting in ghost-like remnants in the data. Typically, motion artifacts in 3D scanning manifest as streaking, blurring, or partial meshes, disrupting an otherwise static environment. As the scanner rotates and collects points, it assumes the scene is stationary; if an object (or person) moves, the collected points for that object are scattered or duplicated along its path of motion.





Orientation and Topology Errors

Algorithmic confusion can flip surface normals (i.e., which side of a mesh is “outside” vs. “inside”). In practice, an inverted mesh might mean a ceiling is rendered as if viewed from above rather than below, or a room appears mirrored (possibly due to glass). This error occurs when software misinterprets the scanner’s position relative to the surface. Although inverted meshes might not be immediately obvious in architectural scans, they can cause problems in analysis and visualization.