Cutting and packing problems involving rectangular-shaped items can be solved optimally with a set of points on which to pack items. Several authors have proposed sets, for example, the normal patterns, the reduced raster points, the regular normal patterns, and the meet-in-the-middle patterns. Many of these sets guarantee an optimal solution to the problem being solved. We investigate the set proposed in 1987, known as reduced raster points, for which we give a formal proof of optimality, demonstrate some properties, propose an efficient algorithm for computing it, and show experiments evidencing their effectiveness over other sets.