Closed Yitzi2 closed 4 years ago
I started looking at it myself, and Perlin noise can actually reach outside these bounds without too much difficulty; at position (0.5, 0.5, 0.25) of a cell, if each corner has one of the gradients that maximizes the dot product, the resulting interpolation yields 17/16, and perturbing y and z slightly can result in an even higher result. (This may be because, unlike standard Perlin noise, the per-node gradients have norm of sqrt(2) rather than 1).
According to the wiki, all outputs are approximately bounded from -1.0 to 1.0, except for distance functions on cellular noise. Are there any exact bounds (even larger ones) on the outputs?