perlinNoise

A Perlin noise generation library implemented in Go

Build

To build a shared library accessible from Python

go build -o perlin_noise.so -buildmode=c-shared main/src

You made need to include the following build flags

CGO_ENABLED=1

As well as operating and platform specific flags. For example on an Apple Silicon Mac,

GOOS=darwin GOARCH=arm64

Tests and Examples

To run perlinNoise and generate example simplex and Perlin noise images,

go run main/src

Or, you can run the Python test script that serves as a demo of how to integrate perlinNoise into your project. Ensure you have the required dependencies:

python3 --version
pip3 --version
pip3 install -r requirements.txt

To run the script,

python3 example.py

Output

perlinNoise currently outputs a matrix of noise of which the values range from 0.0 to 1.0.

Interface and Parameters

Perlin noise can be generated using the generatePerlinNoise function. See test.py for an example of how to load and access the library in Python.

Parameter	Type	Description
resultPtr	*float32	Pointer to the output matrix (faked dimensionality)[1]
width	uint32	Width of resultant matrix.
height	uint32	Height of resultant matrix.
persistence	float32	Intensity falloff coefficient of subsequent noise layers.
numLayers	uint32	Number of simplex noise layers to use.
roughness	float32	Frequency increase coefficient for subsequent noise layers.
baseRoughness	float32	Initial frequency for noise
strength	float32	Scalar multiplier for noise values
randomSeed	float32	Define the random seed to be used

Controlling the noise's behaviour

Behaviour of the noise can be changed by changing parameters to the generatePerlinNoise interface. Settings include width and height to determine the dimensions of the resulting matrix. baseRoughness and roughness to control the frequency and frequency falloff. persistence to change the impact of subsequent layers (low value results in a "softer" look). numLayers controls how many layers of noise will be used (more layers results in more complex, structured noise). strength is a simple scalar multiplier to the matrix (control intensity of noise). Lastly, randomSeed can be changed to change the seed of the noise (perlinNoise is deterministic when randomSeed is known).

Faked Dimensionality

For reduced complexity of the interface between the compiled library and the user, two-dimensionality is faked for the output matrix. Instead of an actual [width, height] matrix, a vector of width * height elements will be used. Nonetheless, we can index what would be at [x, y] if we were using an actual matrix by indexing [x * width + y] in our vector, thus faking the dimensions. The result for the use case of perlinNoise is identical functionality with a simpler interface.

It is easy to restore the actual matrix. Here's an example with NumPy, and without NumPy in Python. Assuming that output_vector is the resultant vector returned by generatePerlinNoise with which was called with width=N and height=M.

# Example using NumPy
import numpy as np
noise_vector: np.ndarray = np.array(output_vector, 'f')
noise_matrix: np.ndarray = noise_vector.reshape(N, M)

# Example without NumPy
noise_matrix = [[], []]
for x in range(N):
    for y in range(M):
        noise_matrix[x, y] = output_vector[x * N + y]