Nondeterministic Programming Methodology
Learn about our RFC process, Open RFC meetings & more.Join in the discussion! »


3.0.14 • Public • Published


npm version Build Status Coverage Status install size

noise noise noise noise noise

Yet another Javascript noise library. Demonstrations here. Currently supports Perlin noise for any arbitrary dimension and Simplex[1-2]. Eventually might support:

  • Simplex[3-4]


npm install tumult --save

The built files are also available on unpkg:

<script src=""></script>


const tumult = require('tumult')
const simplex2 = new tumult.Simplex2('some_seed')
for (let x = 0; x < 10; x++) {
  for (let y = 0; y < 10; y++) {
    console.log(simplex2.gen(/ 64, y / 64))



Object that stores noise constructors. Below is the full list of constructors:

  • tumult.Simplex1
  • tumult.Simplex2
  • tumult.Perlin1
  • tumult.Perlin2
  • tumult.Perlin3
  • tumult.Perlin4
  • tumult.PerlinN

Every constructor has the following signature:


Returns a noise object.


Type: String | Number

Seed to use for shuffling the permutation look-up table. If no value is passed, Math.random() will be used as a seed.


Noise object returned from invoke a noise constructor; all noise objects have the same API:


Re-seeds the permutation look-up table. If a number is passed, it will be converted to a string which will seed the generator. If no string is passed, .seed() defaults to using Math.random()

noise.gen(x, y, z...)

Generates a noise value given the appropriate dimensions (eg. a simplex2 generator should take two arguments, a perlin3 generator should take three arguments, etc.)

noise.octavate(octaves, x, y, z...)

Applies fractal Brownian motion, summing iterations of the noise (# of iterations = octaves). With each successive iteration, the frequency is doubled and the weight is halved.

Note that the generator created by tumult.PerlinN is variadic, meaning you can get Nth dimensional perlin noise by passing N arguments. Note that the gradient lookup table for perlinN isn't optimised, so calling perlinN(x, y) will likely produce less "attractive" noise than perlin2(x, y).

For quickly displaying heightmaps, I highly recommend using terrapaint.



Consider wrapping your function instead:

const tumult = require('tumult')
const simplex2 = new tumult.Simplex2()
const transform = (x, y) => Math.sin(1 / simplex2(x, y))

Takes in a function which will its this bound to noiseGenerator object, meaning you can call gen and octavate using this.gen, etc. This function should take in the dimensions as parameters, and return a value. .transform will return the new transformed noise function. For example, suppose you want a function which will return sin(1/noise(x/32,y/32)), you can do the following:

const tumult = require('tumult')
const simplex2 = new tumult.Simplex2('seed')
const noise = simplex2.transform(function (x, y) {
  return Math.sin(1 / this.gen(x/32, y/32))
for (let i = 0; i < 100; i++) {
  for (let j = 0; j < 100; j++) {
    console.log(noise(i, j))

TL;DR, noise.transform is essentially a helper function that lets you wrap the noise function with your own function.

Note on testing

Currently the tests only verify trivial test requirements (eg. presence of methods, checking if output is within expected [-1, 1] bound); a better way to test this library would be to utilize OpenCV to verify the noise produced is correct, outlined here:

Unfortunately I'm lacking the bandwidth to implement this, but pull requests are welcome!


Perlin noise was invented in 1985 by Ken Perlin.


npm i tumult

DownloadsWeekly Downloads






Unpacked Size

906 kB

Total Files


Last publish


  • avatar