Class TurbulencePatternRed

java.lang.Object
org.apache.batik.ext.awt.image.rendered.AbstractRed
org.apache.batik.ext.awt.image.rendered.TurbulencePatternRed
All Implemented Interfaces:
RenderedImage, CachableRed

public final class TurbulencePatternRed extends AbstractRed
This class creates a RenderedImage in conformance to the one defined for the feTurbulence filter of the SVG specification. What follows is my high-level description of how the noise is generated. This is not contained in the SVG spec, just the algorithm for doing it. This is provided in the hope that someone will figure out a clever way to accelerate parts of the function. gradient contains a long list of random unit vectors. For each point we are to generate noise for we do two things. first we use the latticeSelector to 'co-mingle' the integer portions of x and y (this allows us to have a one-dimensional array of gradients that appears 2 dimensional, by using the co-mingled index). We do this for [x,y], [x+1,y], [x,y+1], and [x+1, y+1], this gives us the four gradient vectors that surround the point (b00, b10, ...) Next we construct the four vectors from the grid points (where the gradient vectors are defined) [these are rx0, rx1, ry0, ry1]. We then take the dot product between the gradient vectors and the grid point vectors (this gives the portion of the grid point vector that projects along the gradient vector for each grid point). These four dot projects are then combined with linear interpolation. The weight factor for the linear combination is the result of applying the 's' curve function to the fractional part of x and y (rx0, ry0). The S curve function get's it's name because it looks a bit like as 'S' from 0->1.
Version:
$Id: TurbulencePatternRed.java 1808001 2017-09-11 09:51:29Z ssteiner $
  • Field Details

    • stitchInfo

      private TurbulencePatternRed.StitchInfo stitchInfo
      Used when stitching is on
    • IDENTITY

      private static final AffineTransform IDENTITY
      Identity transform, default used when null input in the constructor.
    • baseFrequencyX

      private double baseFrequencyX
      x-axis base frequency for the noise function along the x-axis
    • baseFrequencyY

      private double baseFrequencyY
      y-axis base frequency for the noise function along the y-axis
    • numOctaves

      private int numOctaves
      Number of octaves in the noise function
    • seed

      private int seed
      Starting number for the pseudo random number generator
    • tile

      private Rectangle2D tile
      Defines the tile for the turbulence function, if non-null turns on stitching, so frequencies are adjusted to avoid discontinuities in case frequencies do not match tile boundaries.
    • txf

      private AffineTransform txf
      Defines the tile for the turbulence function
    • isFractalNoise

      private boolean isFractalNoise
      Defines whether the filter performs a fractal noise or a turbulence function
    • channels

      private int[] channels
      List of channels that the generator produces.
    • tx

      double[] tx
    • ty

      double[] ty
    • RAND_m

      private static final int RAND_m
      Produces results in the range [1, 2**31 - 2]. Algorithm is: r = (a * r) mod m where a = 16807 and m = 2**31 - 1 = 2147483647 See [Park invalid input: '&' Miller], CACM vol. 31 no. 10 p. 1195, Oct. 1988 To test: the algorithm should produce the result 1043618065 as the 10,000th generated number if the original seed is 1.
      See Also:
    • RAND_a

      private static final int RAND_a
      See Also:
    • RAND_q

      private static final int RAND_q
      See Also:
    • RAND_r

      private static final int RAND_r
      See Also:
    • BSize

      private static final int BSize
      See Also:
    • BM

      private static final int BM
      See Also:
    • PerlinN

      private static final double PerlinN
      See Also:
    • latticeSelector

      private final int[] latticeSelector
    • gradient

      private final double[] gradient
  • Constructor Details

    • TurbulencePatternRed

      public TurbulencePatternRed(double baseFrequencyX, double baseFrequencyY, int numOctaves, int seed, boolean isFractalNoise, Rectangle2D tile, AffineTransform txf, Rectangle devRect, ColorSpace cs, boolean alpha)
      Parameters:
      baseFrequencyX - x-axis base frequency for the noise function along the x-axis
      baseFrequencyY - y-axis base frequency for the noise function along the x-axis
      numOctaves - number of octaves in the noise function. Positive integral value.
      seed - starting number for the pseudo random number generator
      isFractalNoise - defines whether the filter performs a fractal noise or a turbulence function.
      tile - defines the tile size. May be null if stitchTiles is false. Otherwise, should not be null.
      txf - The affine transform from device to user space.
      cs - The Colorspace to output.
      alpha - True if the data should have an alpha channel.
  • Method Details

    • getBaseFrequencyX

      public double getBaseFrequencyX()
    • getBaseFrequencyY

      public double getBaseFrequencyY()
    • getNumOctaves

      public int getNumOctaves()
    • getSeed

      public int getSeed()
    • getTile

      public Rectangle2D getTile()
    • isFractalNoise

      public boolean isFractalNoise()
    • getChannels

      public boolean[] getChannels()
    • setupSeed

      public final int setupSeed(int seed)
    • random

      public final int random(int seed)
    • initLattice

      private void initLattice(int seed)
    • s_curve

      private static final double s_curve(double t)
    • lerp

      private static final double lerp(double t, double a, double b)
    • noise2

      private final void noise2(double[] noise, double vec0, double vec1)
      Generate a pixel of noise corresponding to the point vec0,vec1. See class description for a high level discussion of method. This handles cases where channels invalid input: '<'= 4.
      Parameters:
      noise - The place to put the generated noise.
      vec0 - The X coordiate to generate noise for
      vec1 - The Y coordiate to generate noise for
    • noise2Stitch

      private final void noise2Stitch(double[] noise, double vec0, double vec1, TurbulencePatternRed.StitchInfo stitchInfo)
      This version of the noise function implements stitching. If any of the lattice is on the right or bottom edge, the function uses the the latice on the other side of the tile, i.e., the left or right edge.
      Parameters:
      noise - The place to put the generated noise.
      vec0 - The X coordiate to generate noise for
      vec1 - The Y coordiate to generate noise for
      stitchInfo - The stitching information for the noise function.
    • turbulence_4

      private final int turbulence_4(double pointX, double pointY, double[] fSum)
      This is the heart of the turbulence calculation. It returns 'turbFunctionResult', as defined in the spec. This is special case for 4 bands of output.
      Parameters:
      pointX - x coordinate of the point to process.
      pointY - y coordinate of the point to process.
      fSum - array used to avoid reallocating double array for each pixel
      Returns:
      The ARGB pixel value.
    • turbulence

      private final void turbulence(int[] rgb, double pointX, double pointY, double[] fSum, double[] noise)
      This is the heart of the turbulence calculation. It returns 'turbFunctionResult', as defined in the spec.
      Parameters:
      rgb - array for the four color components
      pointX - x coordinate of the point to process.
      pointY - y coordinate of the point to process.
      fSum - array used to avoid reallocating double array for each pixel
      noise - array used to avoid reallocating double array for each pixel
    • turbulenceStitch

      private final void turbulenceStitch(int[] rgb, double pointX, double pointY, double[] fSum, double[] noise, TurbulencePatternRed.StitchInfo stitchInfo)
      This is the heart of the turbulence calculation. It returns 'turbFunctionResult', as defined in the spec.
      Parameters:
      rgb - array for the four color components
      pointX - x coordinate of the point to process.
      pointY - y coordinate of the point to process.
      fSum - array used to avoid reallocating double array for each pixel
      noise - array used to avoid reallocating double array for each pixel
      stitchInfo - The stitching information for the noise function
    • turbulenceFractal_4

      private final int turbulenceFractal_4(double pointX, double pointY, double[] fSum)
      This is the heart of the turbulence calculation. It returns 'turbFunctionResult', as defined in the spec. This handles the case where we are generating 4 channels of noise.
      Parameters:
      pointX - x coordinate of the point to process.
      pointY - y coordinate of the point to process.
      fSum - array used to avoid reallocating double array for each pixel
      Returns:
      The ARGB pixel
    • turbulenceFractal

      private final void turbulenceFractal(int[] rgb, double pointX, double pointY, double[] fSum, double[] noise)
      This is the heart of the turbulence calculation. It returns 'turbFunctionResult', as defined in the spec.
      Parameters:
      rgb - array for the four color components
      pointX - x coordinate of the point to process.
      pointY - y coordinate of the point to process.
      fSum - array used to avoid reallocating double array for each pixel
      noise - array used to avoid reallocating double array for each pixel
    • turbulenceFractalStitch

      private final void turbulenceFractalStitch(int[] rgb, double pointX, double pointY, double[] fSum, double[] noise, TurbulencePatternRed.StitchInfo stitchInfo)
      This is the heart of the turbulence calculation. It returns 'turbFunctionResult', as defined in the spec.
      Parameters:
      rgb - array for the four color components
      pointX - x coordinate of the point to process.
      pointY - y coordinate of the point to process.
      fSum - array used to avoid reallocating double array for each pixel
      noise - array used to avoid reallocating double array for each pixel
      stitchInfo - The stitching information for the noise function
    • copyData

      public WritableRaster copyData(WritableRaster dest)
      Generates a Perlin noise pattern into dest Raster.
      Parameters:
      dest - Raster to fill with the pattern.