From: wlkngowl@unix.asb.com (Mutatis Mutantdis)
Date: Mon, 25 Mar 1996 17:28:21 +0800
To: wlkngowl@unix.asb.com (Mutatis Mutantdis)
Subject: Noise Sphere Plotter in C++ (was Re: Noise sphere graphical tests of randomness)
Message-ID: <199603250551.AAA22514@unix.asb.com>
MIME-Version: 1.0
Content-Type: text/plain


On Sat, 23 Mar 1996 22:29:52 GMT, I wrote:

>I've got a short program for the PC (w/TPascal source) that plots a 
>noise sphere from a file of (pseudo) random data, if anyone is 
>interested.  Requires a VGA card that handles mode 5Fh (640 x 480, 
[..]
>The source has a brief explanation of what noise spheres are and a 
>reference to the Pickover article the program was based on.

>The source code is not copyrighted, and it would be nice to see it ported to
>other systems, or maybe a portable C version that writes the output to a
>PCX file in RBG(?). [I'm not that fluent in C to write one...]
[..]

BTW, check the ftp.funet.fi site...

Somebody sent me a C++ conversion of it. It compiles fine with BCC 4.02.

Enclosed here (thanks to  "James Pate Williams, Jr." <pate@mindspring.com>)

/*
  NOISESPH.PAS, A Noise Sphere plotter written in Turbo Pascal (23-Mar-96)
  Robert Rothenburg Walking-Owl <WlkngOwl@unix.asb.com>
  No copyright is claimed.  No guarantees made.

  Usage: noisesph file

  This program reads a file of random or pseudo-random data and plots
  a noise sphere of the data. Poor RNGs or sampling methods will show
  clear patterns (definite splotches or spirals).

  The theory behind this is to get a set of 3D polar coordinates from
  the RNG and plot them.  An array is kept of the values, which is
  rotated each time a new byte is read (see the code in the main
  procedure).

  Rather than plot one sphere which can be rotated around any axis,
  it was easier to plot the sphere from three different angles.

  This program is based on a description from the article below.  It
  was proposed as a means of testing pseudo-RNGs:

  Pickover, Clifford A. 1995. "Random number generators: pretty good
	 ones are easy to find."  The Visual Computer (1005) 11:369-377.
*/

#include <conio.h>
#include <fstream.h>
#include <graphics.h>
#include <iostream.h>
#include <math.h>
#include <process.h>

struct Cartesian
{
  double x, y, z;
  unsigned Color;
};

struct Polar
{
  double r, theta, phi;
};

int MidA, MidB, MidC, MidY, Scale;

double ByteToReal(unsigned char b)
{
  /*note that there will be some gaps since we're only using the
	 equivalent of an 8-bit decimal here*/
  return  b / 256.0;
}

//Initialize the graphics screen
int InitScreen(void)
{
  int GraphMode, GraphDriver;

  GraphDriver = VGA;
  GraphMode = VGAHI;
  detectgraph(&GraphDriver, &GraphMode);
  initgraph(&GraphDriver, &GraphMode, "");
  Scale = getmaxx() / 6;
  MidA = Scale, MidB = 3 * Scale, MidC = 5 * Scale;
  MidY = getmaxy() / 2;
  if (MidY < Scale) Scale = MidY;
  return graphresult();
}

unsigned int ScaleColor(double)
{
  return LIGHTGRAY;
}

int Round(double x)
{
  return (int) (x + 0.5);
}

void Plot(struct Cartesian& C)
{
  putpixel(MidA + Round(Scale * C.y), MidY - Round(Scale * C.z), C.Color);
  putpixel(MidB + Round(Scale * C.x), MidY - Round(Scale * C.y), C.Color);
  putpixel(MidC + Round(Scale * C.z), MidY - Round(Scale * C.x), C.Color);
}

void PolarToCartesian(struct Polar P, struct Cartesian& C)
{
  //No rotation was added. Instead we plot from three angles...
  C.x = P.r * sin(P.phi) * cos(P.theta);
  C.y = P.r * sin(P.phi) * sin(P.theta);
  C.z = P.r * cos(P.phi);
  //We can assign colors based on x, y, z, r, theta / pi or phi / (2 * pi)
  C.Color = ScaleColor(C.y);
}

void main(int argc, char* argv[])
{
  double X[3];
  int error, i, n = 0;
  unsigned char byte;
  fstream inp;
  Cartesian C;
  Polar P;

  if (argc != 2)
  {
	 cout << "usage: noisesph filename" << endl;
	 exit(1);
  }
  inp.open(argv[1], ios::binary | ios::in);
  if (!inp)
  {
	 cout << "*error*\could not open input file" << endl;
	 exit(1);
  }
  error = InitScreen();
  if (error != grOk)
  {
	 cout << grapherrormsg(error) << endl;
	 exit(1);
  }
  for (i = 0; i < 3; i++)
  {
	 inp.get(byte);
	 X[i] = ByteToReal(byte);
  }
  do
  {
	 P.r = sqrt(X[(n + 2) % 3]);
	 P.theta = M_PI * X[(n + 1) % 3];
	 P.phi = 2 * M_PI * X[n];
	 PolarToCartesian(P, C);
	 Plot(C);
	 inp.get(byte);
	 X[n] = ByteToReal(byte);
	 n = (n + 1) % 3;
  }
  while (!kbhit() && inp);
  getch();
  closegraph();
}