Terrain Generation

Started by
5 comments, last by Matthewj234 11 years, 8 months ago
Hey guys, I have been frustrated all day on this issue....

I have been using this code to try and generate a 3D environment. My issue... Everything I try with this just ends up with a massive number of random blocks, with no clear placement. The examples I have seen on youtube and elsewhere have shown a coherent noise generation which is procedural. So my question is this, how would I be able to use the code below, to allow me to generate a 3D environment?

Also, why does this only accept [-1,1]?

Thanks in advance for any help offered.


/*****************************************************************************
* J3D.org Copyright (c) 2000
* Java Source
*
* This source is licensed under the GNU LGPL v2.1
* Please read http://www.gnu.org/copyleft/lgpl.html for more information
*
* This software comes with the standard NO WARRANTY disclaimer for any
* purpose. Use it at your own risk. If there's a problem you get to fix it.
*
****************************************************************************/
package com.Sparked_Studios.Tile_Engine.ContentGenerators;
import java.util.Random;
/**
* Computes Perlin Noise for three dimensions.
* <p>
*
* The result is a continuous function that interpolates a smooth path
* along a series random points. The function is consitent, so given
* the same parameters, it will always return the same value. The smoothing
* function is based on the Improving Noise paper presented at Siggraph 2002.
* <p>
* Computing noise for one and two dimensions can make use of the 3D problem
* space by just setting the un-needed dimensions to a fixed value.
*
* @author Justin Couch
* @version $Revision: 1.4 $
*/
public class PerlinNoiseGenerator
{
// Constants for setting up the Perlin-1 noise functions
private static final int B = 0x1000;
private static final int BM = 0xff;
private static final int N = 0x1000;
private static final int NP = 12; /* 2^N */
private static final int NM = 0xfff;
/** Default seed to use for the random number generation */
private static final int DEFAULT_SEED = 100;
/** Default sample size to work with */
private static final int DEFAULT_SAMPLE_SIZE = 256;
/** The log of 1/2 constant. Used Everywhere */
private static final float LOG_HALF = (float)Math.log(0.5);
/** Permutation array for the improved noise function */
private int[] p_imp;
/** P array for perline 1 noise */
private int[] p;
private float[][] g3;
private float[][] g2;
private float[] g1;

/**
* Create a new noise creator with the default seed value
*/
public PerlinNoiseGenerator()
{
this(DEFAULT_SEED);
}
/**
* Create a new noise creator with the given seed value for the randomness
*
* @param seed The seed value to use
*/
public PerlinNoiseGenerator(int seed)
{
p_imp = new int[DEFAULT_SAMPLE_SIZE << 1];
int i, j, k;
Random rand = new Random(seed);
// Calculate the table of psuedo-random coefficients.
for(i = 0; i < DEFAULT_SAMPLE_SIZE; i++)
p_imp = i;
// generate the psuedo-random permutation table.
while(--i > 0)
{
k = p_imp;
j = (int)(rand.nextLong() & DEFAULT_SAMPLE_SIZE);
p_imp = p_imp[j];
p_imp[j] = k;
}
initPerlin1();
}
/**
* Computes noise function for three dimensions at the point (x,y,z).
*
* @param x x dimension parameter
* @param y y dimension parameter
* @param z z dimension parameter
* @return the noise value at the point (x, y, z)
*/
public double improvedNoise(double x, double y, double z)
{
// Constraint the point to a unit cube
int uc_x = (int)Math.floor(x) & 255;
int uc_y = (int)Math.floor(y) & 255;
int uc_z = (int)Math.floor(z) & 255;
// Relative location of the point in the unit cube
double xo = x - Math.floor(x);
double yo = y - Math.floor(y);
double zo = z - Math.floor(z);
// Fade curves for x, y and z
double u = fade(xo);
double v = fade(yo);
double w = fade(zo);
// Generate a hash for each coordinate to find out where in the cube
// it lies.
int a = p_imp[uc_x] + uc_y;
int aa = p_imp[a] + uc_z;
int ab = p_imp[a + 1] + uc_z;
int b = p_imp[uc_x + 1] + uc_y;
int ba = p_imp + uc_z;
int bb = p_imp[b + 1] + uc_z;
// blend results from the 8 corners based on the noise function
double c1 = grad(p_imp[aa], xo, yo, zo);
double c2 = grad(p_imp[ba], xo - 1, yo, zo);
double c3 = grad(p_imp[ab], xo, yo - 1, zo);
double c4 = grad(p_imp[bb], xo - 1, yo - 1, zo);
double c5 = grad(p_imp[aa + 1], xo, yo, zo - 1);
double c6 = grad(p_imp[ba + 1], xo - 1, yo, zo - 1);
double c7 = grad(p_imp[ab + 1], xo, yo - 1, zo - 1);
double c8 = grad(p_imp[bb + 1], xo - 1, yo - 1, zo - 1);
return lerp(w, lerp(v, lerp(u, c1, c2), lerp(u, c3, c4)),
lerp(v, lerp(u, c5, c6), lerp(u, c7, c8)));
}
/**
* 1-D noise generation function using the original perlin algorithm.
*
* @param x Seed for the noise function
* @return The noisy output
*/
public float noise1(float x)
{
float t = x + N;
int bx0 = ((int) t) & BM;
int bx1 = (bx0 + 1) & BM;
float rx0 = t - (int) t;
float rx1 = rx0 - 1;
float sx = sCurve(rx0);
float u = rx0 * g1[p[bx0]];
float v = rx1 * g1[p[bx1]];
return lerp(sx, u, v);
}
/**
* Create noise in a 2D space using the orignal perlin noise algorithm.
*
* @param x The X coordinate of the location to sample
* @param y The Y coordinate of the location to sample
* @return A noisy value at the given position
*/
public float noise2(float x, float y)
{
float t = x + N;
int bx0 = ((int)t) & BM;
int bx1 = (bx0 + 1) & BM;
float rx0 = t - (int)t;
float rx1 = rx0 - 1;
t = y + N;
int by0 = ((int)t) & BM;
int by1 = (by0 + 1) & BM;
float ry0 = t - (int)t;
float ry1 = ry0 - 1;
int i = p[bx0];
int j = p[bx1];
int b00 = p[i + by0];
int b10 = p[j + by0];
int b01 = p[i + by1];
int b11 = p[j + by1];
float sx = sCurve(rx0);
float sy = sCurve(ry0);
float[] q = g2[b00];
float u = rx0 * q[0] + ry0 * q[1];
q = g2[b10];
float v = rx1 * q[0] + ry0 * q[1];
float a = lerp(sx, u, v);
q = g2[b01];
u = rx0 * q[0] + ry1 * q[1];
q = g2[b11];
v = rx1 * q[0] + ry1 * q[1];
float b = lerp(sx, u, v);
return lerp(sy, a, b);
}
/**
* Create noise in a 3D space using the orignal perlin noise algorithm.
*
* @param x The X coordinate of the location to sample
* @param y The Y coordinate of the location to sample
* @param z The Z coordinate of the location to sample
* @return A noisy value at the given position
*/
public float noise3(float x, float y, float z)
{
float t = x + (float)N;
int bx0 = ((int)t) & BM;
int bx1 = (bx0 + 1) & BM;
float rx0 = (float)(t - (int)t);
float rx1 = rx0 - 1;
t = y + (float)N;
int by0 = ((int)t) & BM;
int by1 = (by0 + 1) & BM;
float ry0 = (float)(t - (int)t);
float ry1 = ry0 - 1;
t = z + (float)N;
int bz0 = ((int)t) & BM;
int bz1 = (bz0 + 1) & BM;
float rz0 = (float)(t - (int)t);
float rz1 = rz0 - 1;
int i = p[bx0];
int j = p[bx1];
int b00 = p[i + by0];
int b10 = p[j + by0];
int b01 = p[i + by1];
int b11 = p[j + by1];
t = sCurve(rx0);
float sy = sCurve(ry0);
float sz = sCurve(rz0);
float[] q = g3[b00 + bz0];
float u = (rx0 * q[0] + ry0 * q[1] + rz0 * q[2]);
q = g3[b10 + bz0];
float v = (rx1 * q[0] + ry0 * q[1] + rz0 * q[2]);
float a = lerp(t, u, v);
q = g3[b01 + bz0];
u = (rx0 * q[0] + ry1 * q[1] + rz0 * q[2]);
q = g3[b11 + bz0];
v = (rx1 * q[0] + ry1 * q[1] + rz0 * q[2]);
float b = lerp(t, u, v);
float c = lerp(sy, a, b);
q = g3[b00 + bz1];
u = (rx0 * q[0] + ry0 * q[1] + rz1 * q[2]);
q = g3[b10 + bz1];
v = (rx1 * q[0] + ry0 * q[1] + rz1 * q[2]);
a = lerp(t, u, v);
q = g3[b01 + bz1];
u = (rx0 * q[0] + ry1 * q[1] + rz1 * q[2]);
q = g3[b11 + bz1];
v = (rx1 * q[0] + ry1 * q[1] + rz1 * q[2]);
b = lerp(t, u, v);
float d = lerp(sy, a, b);
return lerp(sz, c, d);
}
/**
* Create a turbulent noise output based on the core noise function. This
* uses the noise as a base function and is suitable for creating clouds,
* marble and explosion effects. For example, a typical marble effect would
* set the colour to be:
* <pre>
* sin(point + turbulence(point) * point.x);
* </pre>
*/
public double imporvedTurbulence(double x,
double y,
double z,
float loF,
float hiF)
{
double p_x = x + 123.456f;
double p_y = y;
double p_z = z;
double t = 0;
double f;
for(f = loF; f < hiF; f *= 2)
{
t += Math.abs(improvedNoise(p_x, p_y, p_z)) / f;
p_x *= 2;
p_y *= 2;
p_z *= 2;
}
return t - 0.3;
}
/**
* Create a turbulance function in 2D using the original perlin noise
* function.
*
* @param x The X coordinate of the location to sample
* @param y The Y coordinate of the location to sample
* @param freq The frequency of the turbluance to create
* @return The value at the given coordinates
*/
public float turbulence2(float x, float y, float freq)
{
float t = 0;
do
{
t += noise2(freq * x, freq * y) / freq;
freq *= 0.5f;
}
while (freq >= 1);
return t;
}
/**
* Create a turbulance function in 3D using the original perlin noise
* function.
*
* @param x The X coordinate of the location to sample
* @param y The Y coordinate of the location to sample
* @param z The Z coordinate of the location to sample
* @param freq The frequency of the turbluance to create
* @return The value at the given coordinates
*/
public float turbulence3(float x, float y, float z, float freq)
{
float t = 0;
do
{
t += noise3(freq * x, freq * y, freq * z) / freq;
freq *= 0.5f;
}
while (freq >= 1);
return t;
}
/**
* Create a 1D tileable noise function for the given width.
*
* @param x The X coordinate to generate the noise for
* @param w The width of the tiled block
* @return The value of the noise at the given coordinate
*/
public float tileableNoise1(float x, float w)
{
return (noise1(x) * (w - x) +
noise1(x - w) * x) / w;
}
/**
* Create a 2D tileable noise function for the given width and height.
*
* @param x The X coordinate to generate the noise for
* @param y The Y coordinate to generate the noise for
* @param w The width of the tiled block
* @param h The height of the tiled block
* @return The value of the noise at the given coordinate
*/
public float tileableNoise2(float x, float y, float w, float h)
{
return (noise2(x, y) * (w - x) * (h - y) +
noise2(x - w, y) * x * (h - y) +
noise2(x, y - h) * (w - x) * y +
noise2(x - w, y - h) * x * y) / (w * h);
}
/**
* Create a 3D tileable noise function for the given width, height and
* depth.
*
* @param x The X coordinate to generate the noise for
* @param y The Y coordinate to generate the noise for
* @param z The Z coordinate to generate the noise for
* @param w The width of the tiled block
* @param h The height of the tiled block
* @param d The depth of the tiled block
* @return The value of the noise at the given coordinate
*/
public float tileableNoise3(float x,
float y,
float z,
float w,
float h,
float d)
{
return (noise3(x, y, z) * (w - x) * (h - y) * (d - z) +
noise3(x - w, y, z) * x * (h - y) * (d - z) +
noise3(x, y - h, z) * (w - x) * y * (d - z) +
noise3(x - w, y - h, z) * x * y * (d - z) +
noise3(x, y, z - d) * (w - x) * (h - y) * z +
noise3(x - w, y, z - d) * x * (h - y) * z +
noise3(x, y - h, z - d) * (w - x) * y * z +
noise3(x - w, y - h, z - d) * x * y * z) /
(w * h * d);
}
/**
* Create a turbulance function that can be tiled across a surface in 2D.
*
* @param x The X coordinate of the location to sample
* @param y The Y coordinate of the location to sample
* @param w The width to tile over
* @param h The height to tile over
* @param freq The frequency of the turbluance to create
* @return The value at the given coordinates
*/
public float tileableTurbulence2(float x,
float y,
float w,
float h,
float freq)
{
float t = 0;
do
{
t += tileableNoise2(freq * x, freq * y, w * freq, h * freq) / freq;
freq *= 0.5f;
}
while (freq >= 1);
return t;
}
/**
* Create a turbulance function that can be tiled across a surface in 3D.
*
* @param x The X coordinate of the location to sample
* @param y The Y coordinate of the location to sample
* @param z The Z coordinate of the location to sample
* @param w The width to tile over
* @param h The height to tile over
* @param d The depth to tile over
* @param freq The frequency of the turbluance to create
* @return The value at the given coordinates
*/
public float tileableTurbulence3(float x,
float y,
float z,
float w,
float h,
float d,
float freq)
{
float t = 0;
do
{
t += tileableNoise3(freq * x,
freq * y,
freq * z,
w * freq,
h * freq,
d * freq) / freq;
freq *= 0.5f;
}
while (freq >= 1);
return t;
}

/**
* Simple lerp function using doubles.
*/
private double lerp(double t, double a, double b)
{
return a + t * (b - a);
}
/**
* Simple lerp function using floats.
*/
private float lerp(float t, float a, float b)
{
return a + t * (b - a);
}
/**
* Fade curve calculation which is 6t^5 - 15t^4 + 10t^3. This is the new
* algorithm, where the old one used to be 3t^2 - 2t^3.
*
* @param t The t parameter to calculate the fade for
* @return the drop-off amount.
*/
private double fade(double t)
{
return t * t * t * (t * (t * 6 - 15) + 10);
}
/**
* Calculate the gradient function based on the hash code.
*/
private double grad(int hash, double x, double y, double z)
{
// Convert low 4 bits of hash code into 12 gradient directions.
int h = hash & 15;
double u = (h < 8 || h == 12 || h == 13) ? x : y;
double v = (h < 4 || h == 12 || h == 13) ? y : z;
return ((h & 1) == 0 ? u : -u) + ((h & 2) == 0 ? v : -v);
}
/**
* Simple bias generator using exponents.
*/
private float bias(float a, float b)
{
return (float)Math.pow(a, Math.log(b) / LOG_HALF);
}

/*
* Gain generator that caps to the range of [0, 1].
*/
private float gain(float a, float b)
{
if(a < 0.001f)
return 0;
else if (a > 0.999f)
return 1.0f;
double p = Math.log(1.0f - b) / LOG_HALF;
if(a < 0.5f)
return (float)(Math.pow(2 * a, p) / 2);
else
return 1 - (float)(Math.pow(2 * (1.0f - a), p) / 2);
}
/**
* S-curve function for value distribution for Perlin-1 noise function.
*/
private float sCurve(float t)
{
return (t * t * (3 - 2 * t));
}
/**
* 2D-vector normalisation function.
*/
private void normalize2(float[] v)
{
float s = (float)(1 / Math.sqrt(v[0] * v[0] + v[1] * v[1]));
v[0] *= s;
v[1] *= s;
}
/**
* 3D-vector normalisation function.
*/
private void normalize3(float[] v)
{
float s = (float)(1 / Math.sqrt(v[0] * v[0] + v[1] * v[1] + v[2] * v[2]));
v[0] *= s;
v[1] *= s;
v[2] *= s;
}
/**
* Initialise the lookup arrays used by Perlin 1 function.
*/
private void initPerlin1()
{
p = new int[B + B + 2];
g3 = new float[B + B + 2][3];
g2 = new float[B + B + 2][2];
g1 = new float[B + B + 2];
int i, j, k;
for(i = 0; i < B; i++)
{
p = i;
g1 = (float)(((Math.random() * Integer.MAX_VALUE) % (B + B)) - B) / B;
for(j = 0; j < 2; j++)
g2[j] = (float)(((Math.random() * Integer.MAX_VALUE) % (B + B)) - B) / B;
normalize2(g2);
for(j = 0; j < 3; j++)
g3[j] = (float)(((Math.random() * Integer.MAX_VALUE) % (B + B)) - B) / B;
normalize3(g3);
}
while(--i > 0)
{
k = p;
j = (int)((Math.random() * Integer.MAX_VALUE) % B);
p = p[j];
p[j] = k;
}
for(i = 0; i < B + 2; i++)
{
p[B + i] = p;
g1[B + i] = g1;
for(j = 0; j < 2; j++)
g2[B + i][j] = g2[j];
for(j = 0; j < 3; j++)
g3[B + i][j] = g3[j];
}
}
}



My current implementation.

public void initChunk(PerlinNoiseGenerator png) {

for (int x = 0; x < 16; x++) {
for (int z = 0; z < 16; z++) {
for (int y = 0; y < 256; y++) {
double noise = ImprovedNoise.noise(x*0.4, y*0.4, z*0.4);
//log(noise);
if (noise < 0) noise = -noise;
if (noise >0.5) chunkVoxels[x][y][z] = new VoxelRef(x + xPos, y, z + zPos, 1, 1, 1, 1);
}
}
}
}

- Numprt

Advertisement
One common error wit Perlin noise is that it looks random if you sample it at integer values. The smoothing occurs between integers, so I would suggest effectively zooming in.
I would highly recommend using LibNoise over rolling your own noise algorithm. You'll get more consistent results and you'll have access to many other extremely useful features.
Also, for generating height maps, you can use the plasma fractal algorithm.
See here:
http://en.wikipedia.org/wiki/Diamond-square_algorithm

I would highly recommend using LibNoise over rolling your own noise algorithm. You'll get more consistent results and you'll have access to many other extremely useful features.


Thanks but Im using java. I had looked at this, and im trying to find a java port. However, I believe that the perlin code is correct, it is from J3D, however my method of implementation is whats causing the issues.


One common error wit Perlin noise is that it looks random if you sample it at integer values. The smoothing occurs between integers, so I would suggest effectively zooming in.


I sample at 1-16*0.4, so I believe all of my samples are doubles.

- Numprt

It's not about whether they are integers, but how many samples you take between integers. The way Perlin noise is calculated, the values at integer co-ordinates are pseudo-random. If you only sampled at integer values (e.g. (1.0, 2.0)) the values would look totally random. If you went by steps of 0.5, the values in between would be a combination of the values from the integer co-ordinates (e.g. the value at (1.5, 2.0) would be part way between the value at (1.0, 2.0) and (2.0, 2.0)). The smaller the steps you take, the smoother the values will look. I would suggest finer steps, e.g. 0.1 instead of 0.4.

It's not about whether they are integers, but how many samples you take between integers. The way Perlin noise is calculated, the values at integer co-ordinates are pseudo-random. If you only sampled at integer values (e.g. (1.0, 2.0)) the values would look totally random. If you went by steps of 0.5, the values in between would be a combination of the values from the integer co-ordinates (e.g. the value at (1.5, 2.0) would be part way between the value at (1.0, 2.0) and (2.0, 2.0)). The smaller the steps you take, the smoother the values will look. I would suggest finer steps, e.g. 0.1 instead of 0.4.


Thanks, I am now using 0.01 steps and the results are AMAZING! So thank you very much!



Thanks everyone, you helped alot!

- Numprt

This topic is closed to new replies.

Advertisement