#region Copyright & License Information
/*
* Copyright (c) The OpenRA Developers and Contributors
* This file is part of OpenRA, which is free software. It is made
* available to you under the terms of the GNU General Public License
* as published by the Free Software Foundation, either version 3 of
* the License, or (at your option) any later version. For more
* information, see COPYING.
*/
#endregion
using System;
using System.Numerics;
using OpenRA.Support;
namespace OpenRA.Mods.Common.MapGenerator
{
public static class NoiseUtils
{
const int Scale = 1024;
const int ScaledSqrt2 = 1448;
/// Amplitude is the same for all wavelengths.
public static int WhiteAmplitude(int wavelength) => 1;
/// Amplitude proportional to wavelength.
public static int PinkAmplitude(int wavelength) => wavelength;
///
/// amplitude = wavelength ** (1 / (2 ** clumpiness))
/// Setting clumpiness to 0 is equivalent to pink noise.
///
public static int ClumpinessAmplitude(int wavelength, int clumpiness)
{
var amplitude = wavelength;
for (var i = 0; i < clumpiness; i++)
amplitude = Exts.ISqrt(amplitude);
return amplitude;
}
///
///
/// Create noise by combining multiple layers of Perlin noise of halving wavelengths.
///
///
/// featureSize defines the largest wavelength in 1024ths of a matrix cell.
/// the output.
///
///
/// ampFunc specifies the amplitude of each wavelength. PinkAmplitude is often a suitable
/// choice.
///
///
public static Matrix FractalNoise(
MersenneTwister random,
int2 size,
int featureSize,
Func ampFunc)
{
var span = Math.Max(size.X, size.Y);
var wavelengths = new int[BitOperations.Log2((uint)span)];
for (var i = 0; i < wavelengths.Length; i++)
wavelengths[i] = featureSize >> i;
var noise = new Matrix(size);
foreach (var wavelength in wavelengths)
{
if (wavelength <= Scale / 2)
break;
var amps = ampFunc(wavelength);
var subSpan = span * Scale / wavelength + 2;
var subNoise = PerlinNoise(random, subSpan);
// Offsets should align to grid.
// (The wavelength is divided back out later.)
var scaledOffsetX = (int)(random.NextUint() % (wavelength + 1));
var scaledOffsetY = (int)(random.NextUint() % (wavelength + 1));
for (var y = 0; y < size.Y; y++)
for (var x = 0; x < size.X; x++)
{
var scaledMappedX = x * Scale + scaledOffsetX;
var scaledMappedY = y * Scale + scaledOffsetY;
noise[y * size.X + x] +=
amps * MatrixUtils.IntegerInterpolate(
subNoise,
scaledMappedX / wavelength,
scaledMappedY / wavelength,
scaledMappedX % wavelength,
scaledMappedY % wavelength,
wavelength);
}
}
return noise;
}
///
/// 2D Perlin Noise generator without interpolation, producing a span-by-span sized matrix.
/// Output values range from -5792 to +5792.
///
public static Matrix PerlinNoise(MersenneTwister random, int span)
{
var noise = new Matrix(span, span);
for (var y = 0; y <= span; y++)
for (var x = 0; x <= span; x++)
{
var phase = new WAngle((int)random.NextUint() % 1024);
var vx = phase.Cos();
var vy = phase.Sin();
if (x > 0 && y > 0)
noise[x - 1, y - 1] += -vx + -vy;
if (x < span && y > 0)
noise[x, y - 1] += vx + -vy;
if (x > 0 && y < span)
noise[x - 1, y] += -vx + vy;
if (x < span && y < span)
noise[x, y] += vx + vy;
}
return noise;
}
///
///
/// Produce symmetric 2D noise by repeatedly applying some generated Perlin noise under
/// rotation and mirroring.
///
///
/// Note that the combination of multiple noise values with varying correlations creates a
/// noise with different properties to simple Perlin noise.
///
///
public static Matrix SymmetricFractalNoise(
MersenneTwister random,
int2 size,
int rotations,
Symmetry.Mirror mirror,
int featureSize,
Func ampFunc)
{
if (rotations < 1)
throw new ArgumentException("rotations must be >= 1");
// Need higher resolution due to cropping and rotation artifacts
var templateSpan = Math.Max(size.X, size.Y) * 2 + 2;
var templateSize = new int2(templateSpan, templateSpan);
var scaledTemplateCenter = new int2(templateSpan - 1, templateSpan - 1) * Scale / 2;
var template = FractalNoise(random, templateSize, featureSize, ampFunc);
var output = new Matrix(size);
var inclusiveOutputSize = size - new int2(1, 1);
var scaledOutputMid = inclusiveOutputSize * Scale / 2;
for (var y = 0; y < size.Y; y++)
for (var x = 0; x < size.X; x++)
{
var outputXy = new int2(x, y);
var scaledOutputXy = outputXy * Scale;
var scaledOutputXyFromCenter = scaledOutputXy - scaledOutputMid;
// Apply sqrt2 scaling so that diagonal samples don't alias.
var scaledTemplateXyFromCenter = scaledOutputXyFromCenter * ScaledSqrt2 / Scale;
var scaledTemplateXy = scaledTemplateXyFromCenter + scaledTemplateCenter;
var projections = Symmetry.RotateAndMirrorPointAround(
scaledTemplateXy, scaledTemplateCenter, rotations, mirror);
foreach (var projection in projections)
output[x, y] +=
MatrixUtils.IntegerInterpolate(
template,
projection.X / Scale,
projection.Y / Scale,
projection.X % Scale,
projection.Y % Scale,
Scale);
}
return output;
}
///
/// Use SymmetricFractalNoise to fill a CellLayer. The noise is aligned to the CPos
/// coordinate system.
///
public static void SymmetricFractalNoiseIntoCellLayer(
MersenneTwister random,
CellLayer cellLayer,
int rotations,
Symmetry.WMirror wmirror,
int featureSize,
Func ampFunc)
{
var cellBounds = CellLayerUtils.CellBounds(cellLayer);
var size = new int2(cellBounds.Size.Width, cellBounds.Size.Height);
var noise = SymmetricFractalNoise(
random,
size,
rotations,
wmirror.ForCPos(),
featureSize,
ampFunc);
CellLayerUtils.FromMatrix(cellLayer, noise);
}
}
}