Viewpoint specification is fundamental to traditional computer graphics rendering. Both the transformation of a scene to eye space in the traditional graphics pipeline and the origination of viewing rays in a ray-casting system depend on the viewpoint. Moreover, many subsequent rendering steps are also impacted by the choice of viewpoint, including clipping, projection, illumination calculations, shading, and visibility determination. As a result, changing the viewpoint frequently gates the entire process of interactive rendering, as each rendered frame is initiated with the specification of a viewpoint, followed by the scene description, and culminating with the final displayed image.

There are many potential advantages to decoupling viewpoint specification from rendering. First, immediate efficiency improvements are available if rendering costs are amortized over multiple views. They result from reuse of shading calculations as well as exploiting the coherency of surface reflection with smooth variations in viewpoint. A second advantage results from beginning the rendering process before the viewing position is resolved, thereby reducing latency.

Download PixelView Tutorial: A View-Independent Graphics Rendering Architecture