Noise engines

Noise engines provide 'noise', pseudo random values for each point or pixel. These are the starting points for creating a terrain.

Noise icons are square and have a red circle at the bottom. A red circle denotes an output channel, so noise engines are output only.

For further information on the noise types available, click on the image map below.

Cached Worley.

CacheWorley implement's a Voronni pattern. An area is covered in random points, then the area is coloured by its closeness to the nearest point
(or second nearest, or third etc). This produces a cellular pattern.

By using different ways to calculate the closeness to a point , which pG calls metrics, pG extends the types of pattern this produces.
The following images use the same Worley parameters but different metrics.

The metrics used, from left to right are are -

Euclidean.: The straight line distance.
Manhattan.: Also called City Block distance or rectilinear distance.
The distance between two points if you could only travel parallel to the axes.
Chessboard.: The highest of the differences between x, y and z coordinates of both points.
SineSquared.: The Euclidean distance squared and then feed into a sine function.

Constant Value.

This produces a constant value for every point. It does not have much use on its own, however it becomes very useful when used with a combiner.

Fault Noise.

Fault Noise has a very simple premise. Create a random plane through 3D space. Raise every point on one side of the plane, decrease every point on the other side. Repeat lots of times. Works best with meshes, especially planets.

Linear Gradient.

Creates a linear gradient between two points. Like Constant Value it is really for use with functions or combiners.

Perlin's Noise.

Ken Perlin's classic, with turbulence. If you are mathematically inclined here's a better explaination, but here is my simplified version

Imagine a 256x256 grid of pseudo random values. Write a function the takes x and y coordinates within the grid and returns the a value interpolated from the surrounding points on the grid.

You calculate the value for a point

result = noise(x,y)

then add to the result

result = result + 1/2 * noise(x*2,y*2)

then add to the result

result = result + 1/4 * noise(x*4,y*4)

and you get 2D Perlin's noise, Extend this to a 256x256x256 cube for 3D Perlin's noise.

Have a play with the values. Its a quick way to nice landscapes. Run it through the absolute function too, this is often called Ridged Perlin's noise, it gives a river network effect, or ridges if you invert it.