POV-Ray : Newsgroups : povray.binaries.images : Re: The Cubic Mandelbrot [23 kb] Server Time
18 Nov 2024 20:23:39 EST (-0500)
  Re: The Cubic Mandelbrot [23 kb] (Message 1 to 6 of 6)  
From: Mike Williams
Subject: Re: The Cubic Mandelbrot [23 kb]
Date: 15 Jun 2005 06:49:07
Message: <42b007a3@news.povray.org>
Wasn't it Orchid XP v2 who wrote:
>Also of interest: slice the isosurface and stick a matching 2D plot on
>the cross-section! (I will do this later if I figure out pigment
>functions...)

Here are cross sections through the three axes. I used 12 iterations here
(pigment functions are a lot faster than 3d isosurfaces).

Here's the source code for the Z axis slice:-

// Just useful.
#declare ReMul = function (Xr, Xi, Yr, Yi) {Xr * Yr - Xi * Yi};
#declare ImMul = function (Xr, Xi, Yr, Yi) {Xr * Yi + Xi * Yr};

#declare Re2 = function (Zr, Zi) {  Zr*Zr - Zi*Zi};
#declare Im2 = function (Zr, Zi) {2*Zr*Zi};

#declare Re3 = function (Zr, Zi) {  Zr*Zr*Zr - 3*Zr*Zi*Zi};
#declare Im3 = function (Zr, Zi) {3*Zr*Zr*Zi -   Zi*Zi*Zi};

// Change if you like...
#declare End = function (Zr, Zi) {Zr*Zr + Zi*Zi};

// Build recursive definition...
#declare Fn = array[21];

#declare Fn[0] = function (Zr, Zi, Tr, Ti, Br, Bi) {End(Zr, Zi)};

#declare lp=1;
#while (lp<21)
   #declare Fn[lp] = function (Zr, Zi, Tr, Ti, Br, Bi)
   {
     Fn[lp-1]
     (
       Re3(Zr, Zi) - ReMul(Zr, Zi, Tr, Ti) + Br,
       Im3(Zr, Zi) - ImMul(Zr, Zi, Tr, Ti) + Bi,
       Tr, Ti, Br, Bi
     )
   };

   #declare lp = lp + 1;
#end
#undef lp

// Change this at will... (Must be < 20 tho.)
#declare Iterations = 12;

// Don't touch...
#declare InitP = function (Ar, Ai, Br, Bi) {Fn[Iterations](+Ar, +Ai,
3*Re2(Ar, Ai), 3*Im2(Ar, Ai), Br, Bi)};
#declare InitM = function (Ar, Ai, Br, Bi) {Fn[Iterations](-Ar, -Ai,
3*Re2(Ar, Ai), 3*Im2(Ar, Ai), Br, Bi)};

// Now draw something!!
camera
{
   location <0, 0, -4.5>
   look_at <0, 0, 0>
}

plane {z,0
  pigment {
    function {min(InitP(z, 0, x, y) - 4, 10)}
    colour_map {
      [0    rgb <0,0,0>]
      [0.0001 rgb <0,0,1>]
      [0.33 rgb <1,1,0>]
      [0.67 rgb <1,0,0>]
      [1.0  rgb <0,0,1>]
    }
  }
  finish {ambient 1}
  rotate z*90
  scale 2
}


Post a reply to this message


Attachments:
Download '2DMandelY.jpg' (28 KB) Download '2DMandelX.jpg' (37 KB) Download '2DMandel.jpg' (42 KB)

Preview of image '2DMandelY.jpg'
2DMandelY.jpg

Preview of image '2DMandelX.jpg'
2DMandelX.jpg

Preview of image '2DMandel.jpg'
2DMandel.jpg


 

From: Stephen McAvoy
Subject: Re: The Cubic Mandelbrot [23 kb]
Date: 15 Jun 2005 10:05:02
Message: <bcd0b11b2umvvdco2jitjhagmreugvef7h@4ax.com>
On Wed, 15 Jun 2005 11:48:59 +0100, "Mike Williams"
<nos### [at] econymdemoncouk> wrote:

>Here are cross sections through the three axes. I used 12 iterations here
>(pigment functions are a lot faster than 3d isosurfaces).
>
>Here's the source code for the Z axis slice:-

It doesn't colour cycle :-)

Regards
        Stephen


Post a reply to this message

From: Orchid XP v2
Subject: Re: The Cubic Mandelbrot [23 kb]
Date: 15 Jun 2005 15:50:15
Message: <42b08677$1@news.povray.org>
>>Also of interest: slice the isosurface and stick a matching 2D plot on
>>the cross-section! (I will do this later if I figure out pigment
>>functions...)
> 
> 
> Here are cross sections through the three axes. I used 12 iterations here
> (pigment functions are a lot faster than 3d isosurfaces).

Neat!

The bottom image looks like one would expect. The other two look like I 
maybe messed up my maths somewhere... they don't look right somehow.

The bottom image I'm guessing is A=(0, 0), B=(x, y). That looks correct. 
If you try A=(x, y), B=(0, 0) it should look like this:

http://www.felicite-parmentier.freeserve.co.uk/large/cubic-mand.gif

However, it looks like the second image is trying to be that, but messed 
up. Looks like I need to go recheck my math! (You'll notice that the 
image that looks right doesn't use A at all - and there's an 
optimisation involved with the A variable. Wanna bet I optimised it wrong?)

Anyways, usually when rendered 2D, the part of the isosurface that's 
solid is coloured black, and the parts outside are multicoloured. ;-) Of 
course, there's no *law* about that...

I'll go see if I can figure out if/where my code is wrong. :-S


Post a reply to this message

From: Mike Williams
Subject: Re: The Cubic Mandelbrot [23 kb]
Date: 15 Jun 2005 16:26:11
Message: <6$VdAAAb7IsCFw2x@econym.demon.co.uk>
Wasn't it Orchid XP v2 who wrote:
>>>Also of interest: slice the isosurface and stick a matching 2D plot on
>>>the cross-section! (I will do this later if I figure out pigment
>>>functions...)
>> 
>> 
>> Here are cross sections through the three axes. I used 12 iterations here
>> (pigment functions are a lot faster than 3d isosurfaces).
>
>Neat!
>
>The bottom image looks like one would expect. The other two look like I 
>maybe messed up my maths somewhere... they don't look right somehow.
>
>The bottom image I'm guessing is A=(0, 0), B=(x, y). That looks correct. 
>If you try A=(x, y), B=(0, 0) it should look like this:
>
>http://www.felicite-parmentier.freeserve.co.uk/large/cubic-mand.gif
>
>However, it looks like the second image is trying to be that, but messed 
>up. Looks like I need to go recheck my math! (You'll notice that the 
>image that looks right doesn't use A at all - and there's an 
>optimisation involved with the A variable. Wanna bet I optimised it wrong?)
>
>Anyways, usually when rendered 2D, the part of the isosurface that's 
>solid is coloured black, and the parts outside are multicoloured. ;-) Of 
>course, there's no *law* about that...
>
>I'll go see if I can figure out if/where my code is wrong. :-S


I suspect that it has something to do with the values of T1 and T2.

When A=(0,0), T1=0 and T2=0, and the whole thing behaves exactly like a
conventional 2D cubic Mandlebrot (z = z^3 + #pixel as they say in an
UltraFractal script).

-- 
Mike Williams
Gentleman of Leisure


Post a reply to this message

From: Emerald Orchid
Subject: Re: The Cubic Mandelbrot [23 kb]
Date: 15 Jun 2005 16:46:13
Message: <42b09395$1@news.povray.org>
>>The bottom image looks like one would expect. The other two look like I 
>>maybe messed up my maths somewhere... they don't look right somehow.
>>
>>I'll go see if I can figure out if/where my code is wrong. :-S
> 
> I suspect that it has something to do with the values of T1 and T2.

Indeed - that's the optimisation.

The formula is z^3 - 3a^2z + b, as - as an optimisation - my code
calculates t = 3a^2. Then the formula is z^3 - tz + b.

Still... I'm staring at my code... and I'm not seeing anything wrong...
hmmmm... 0:-)


Post a reply to this message

From: Mike Williams
Subject: Re: The Cubic Mandelbrot [23 kb]
Date: 15 Jun 2005 17:05:02
Message: <jv40VAAufJsCFwlO@econym.demon.co.uk>
Wasn't it Emerald Orchid who wrote:
>>>The bottom image looks like one would expect. The other two look like I 
>>>maybe messed up my maths somewhere... they don't look right somehow.
>>>
>>>I'll go see if I can figure out if/where my code is wrong. :-S
>> 
>> I suspect that it has something to do with the values of T1 and T2.
>
>Indeed - that's the optimisation.
>
>The formula is z^3 - 3a^2z + b, as - as an optimisation - my code
>calculates t = 3a^2. Then the formula is z^3 - tz + b.
>
>Still... I'm staring at my code... and I'm not seeing anything wrong...
>hmmmm... 0:-)


I just spotted that the formula for the image you want it to look like
appears in the title bar of the gif. Namely:

        Z = Z^3 - 3.c^2.Z + 0

However, the function you're using for the 3D mandelbrot is

        Z = Z^3 - 3.c^2.Z + #pixel

on the X plane cross section, the real part of the #pixel is 0

This code plots Z = Z^3 - 3.c^2.Z + 0 using something similar to your
original 3D function, but I can't see how to get something 3D that does
this on the X axis and plots Z = Z^3 + #pixel on the Z axis at the same
time.


// Just useful.
#declare ReMul = function (Xr, Xi, Yr, Yi) {Xr * Yr - Xi * Yi};
#declare ImMul = function (Xr, Xi, Yr, Yi) {Xr * Yi + Xi * Yr};

#declare Re2 = function (Zr, Zi) {  Zr*Zr - Zi*Zi};
#declare Im2 = function (Zr, Zi) {2*Zr*Zi};

#declare Re3 = function (Zr, Zi) {  Zr*Zr*Zr - 3*Zr*Zi*Zi};
#declare Im3 = function (Zr, Zi) {3*Zr*Zr*Zi -   Zi*Zi*Zi};

// Change if you like...
#declare End = function (Zr, Zi) {Zr*Zr + Zi*Zi};

// Build recursive definition...
#declare Fn = array[21];

#declare Fn[0] = function (Zr, Zi, Tr, Ti, Br, Bi) {End(Zr, Zi)};

#declare lp=1;
#while (lp<21)
   #declare Fn[lp] = function (Zr, Zi, Tr, Ti, Br, Bi)
   {
     Fn[lp-1]
     (
       Re3(Zr, Zi) - ReMul(Zr, Zi, Tr, Ti) + Br,
       Im3(Zr, Zi) - ImMul(Zr, Zi, Tr, Ti) + Bi,
       Tr, Ti, Br, Bi
     )
   };

   #declare lp = lp + 1;
#end
#undef lp

// Change this at will... (Must be < 20 tho.)
#declare Iterations = 12;

// Don't touch...
#declare InitP = function (Ar, Ai, Br, Bi) {Fn[Iterations](+Ar, +Ai, 
3*Re2(Ar, Ai), 3*Im2(Ar, Ai), Br, Bi)};
#declare InitM = function (Ar, Ai, Br, Bi) {Fn[Iterations](-Ar, -Ai, 
3*Re2(Ar, Ai), 3*Im2(Ar, Ai), Br, Bi)};

// Now draw something!!
camera
{
   location <0, 0, -4.5>
   look_at <0, 0, 0>
}

light_source {<-2, +3, -5>, colour rgb <1.0, 0.0, 1.0>}
light_source {<+2, -3, -5>, colour rgb <0.0, 1.0, 0.0>}


plane {x,0
  pigment {
    function {min(InitP(z, y, x, 0) - 4, 10)} // changed this 
    colour_map {
      [0    rgb <0,0,0>]
      [0.0001 rgb <0,0,1>]
      [0.33 rgb <1,1,0>]
      [0.67 rgb <1,0,0>]
      [1.0  rgb <0,0,1>]
    }
  }
  finish {ambient 1}
  scale 2
  rotate <0,90,90>
}
    




-- 
Mike Williams
Gentleman of Leisure


Post a reply to this message

Copyright 2003-2023 Persistence of Vision Raytracer Pty. Ltd.