While it is useful in some contexts (particularly in 3D visualization) to use raw 3D images (e.g., 3D images captured, generated or stored by the computing devices of FIGS. 1 and 2), there can arise compatibility, data alignment or interpolation issues that arise when attempting to use the same raw 3D images in other contexts, for example, when attempting to use the raw 3D image with training or executing predictive models built from machine learning algorithms. In such contexts, for example, the unstructured 3D point cloud data of one 3D image (e.g., stored in a PLY file) could be misaligned with respect to the 3D point cloud data of another 3D image (e.g., stored in another PLY file). For example, if the first point of one raw PLY file represents a point identifying the head of a person, the first point of another raw PLY file could represent a point identifying a hand or a leg. This can create an issue because no meaningful connection can be made between the two 3D images with their differing ordering or arrangement of 3D points when training or executing predictive models with respect to such features.
