3D is a poor choice for many visualizations because when it is rendered in 2D, information is hidden behind the model being transformed over time (either manual

3D is a poor choice for many visualizations because when it is rendered in 2D, information is hidden behind the model being transformed over time (either manually by the user or via automatic rotation, etc). IMO it should only be considered if there are challenges with flattening the data in question. For example, if the data represents an actual scanned 3D object. On the other hand, rendering arbitrary data as "3D" models is usually a cute waste of time. There are much more effective ways to explore and analyze it.