Constructing a 3D visualization is a process of describing the " scene" that will be rendered. This includes the location of the viewing point (i.e., where the "camera" is placed), the method by which a system would be viewed (i.e., the "lens," which may be orthographic, perspective, fisheye, spherical, and so on) and the components that will be rendered (render "sources," such as volume elements, lines, annotations, and opaque surfaces). The 3D plotting infrastructure then develops a resultant image from this scene, which can be saved to a file or viewed inline.

By constructing the scene in this programmatic way, full control can be had over each component in the scene as well as the method by which the scene is rendered; this can be used to prototype visualizations, inject annotation such as a grid or continent lines, and then render a production-quality visualization. By changing the "lens" used, a single camera path can output images suitable for planetarium domes, immersive and head tracking systems (such as the Oculus Rift or recent 360-degree/virtual reality movie viewers such as the mobile YouTube app), as well as standard screens.