Point based graphics learning notes (二)

A common set of challenges:
1) how to edit the scene by manipulating its samples
2) how to store and compress these samples
3) how to transform and shade them
4) how to render them with correct sampling, visibility, and filtering

Image-order algorithm vs. object-order algorithm
Visibility is tantamount to sorting. The gather strategy leads to an image-order algorithm. By contrast, the scatter strategy leads to an object-order algorithm.

Procedural modeling: the generation of scene geometry using a computer algorithm (rather than interactively or by sensing).

To render a procedurally defined object using favored image-order algorithms, one must be able to compute for a given pixel which part of the object (if any) lands there. If the procedure is expensive to invert in this sense, or even uninvertible, then an object-order algorithm rendering must be used.

The curious case of displacement mapping.

The notion of using points and an object-order rendering algorithm to display smooth primitives was displacement-mapped surfaces by Levoy and Whitted.

By the time point-based rendering was resurrected, the graphics landscape had changed considerably. The number of pixels covered by a typical polygon had been shrinking for a decade, and hardware antialiasing was becoming commonplace. These developments mad connectivity less important than it was in 1984, and they made antialiased points a more attractive primitive.