We present a method for generating well-distributed point according to a given
density. Our method is based on a single precomputed tile. The tile is both
self-similar and hierarchical. A selfsimilar tile allows to increase the density
of points in large steps by recursively subdividing the tile. A hierarchical
tile allows to smoothly adjust the density of points. We present an interesting
method to construct a self-similar point distribution, we show how to construct
a well-distributed self-similar point distribution, and how to make a point
distribution hierarchical. Our method is capable of generating well-distributed
point sets in real time, using an algorithm that is easy to implement. However,
because only a single tile is used, noticeable periodicity is introduced in the
generated point distributions. Therefore, our method is somewhat better suited
for applications that do not require point distributions with a high visual
quality, such as sampling.