Print Page - Spherical Texture Fold

Title: Spherical Texture Fold
Post by: Aexion on July 12, 2011, 01:47:13 PM

Hello,

Here is a little idea that I got while coding Incendia bump maps:

The classic mandelbox consist in an box fold and a sphere fold.
I have tweaked a little the sphere fold so in every iteration it can modify the folding by sampling a bump map.
This was done by transforming the actual point into spherical coordinates and converting it into u,v coordinates for a bump map.
Then, the value of the bump map is multiplied by the actual sphere-fold radius calculation.
This procedure texture the sphere walls of the box. It needs optimizations (such as texture interpolations and a better optimized optimization), but it lets you to decorate box interiors with what you have on the texture map.
Obviously, sphere texture maps works better than square ones, but they are hard to find.

Anyways, here is a simple example:
BMap is a bump map array stored in the memory, it is unsigned char and has a (BSizeX, BSizeY) size
The strength factor is 0.05

Code:

bool Texture_MBox(float x, float y, float z,int Iter){
//do not start with (0,0,0) because it will give error on the bump map calculation
const float nx=x;
const float ny=y;
const float nz=z;
int ic;
bool escape=false;

for (ic=0;ic<Iter;ic++){
	if(x>1)x=2-x;
	else 
	if(x<-1)x=-2-x;
	
	if(y>1)y=2-y;
	else 
	if(y<-1)y=-2-y;

	if(z>1)z=2-z;
	else 
	if(z<-1)z=-2-z;

//Spherical Texture Fold
 const float mag = 1/sqrt(x*x+y*y+z*z);   
 const float vx = x*mag;
 const float vy = y*mag;
 const float vz = z*mag;
 const int fxpos=((atan2(vx,vz)/3.14159265358979323846)*0.5+0.5)*BSizeX;//texture width
 const int fypos=((asin(vy)/1.5707963267948966192313216916398)*0.5+0.5)*BSizeY; //texture height
 const float bvalue=(float)(BMap[BSizeX*fypos+fxpos])*0.003921568627450980392156862745098; //(texture data)*(1/255)
 const float vm=sqrt(x*x+y*y+z*z)*(1+bvalue*0.05); //strength=0.05

//const float vm=sqrt(x*x+y*y+z*z); //classic mandel box

if(vm<0.05){
x*=4;
y*=4;
z*=4;
} else 
if(vm<1){
const float vsq=vm*vm;
x/=vsq;
y/=vsq;
z/=vsq;
}
x=x*2+nx;
y=y*2+ny;
z=z*2+nz;
if ((x*x+y*y+z*z)>3600) break;	
	}	
if(ic<Iter)escape=false;else escape=true;
return escape;
}

I have include two very crude renders using a low resolution skull texture and another with a NASA spherical moon texture.

ps. Zooming in a skull decorated mandelbox its a little creepy, better to decorate it with some Egyptian hieroglyphs.. :)

Title: Re: Spherical Texture Fold
Post by: Jesse on July 13, 2011, 12:28:33 AM

That is really cool!

I have to add a pointer for formulas to call a map function,
and a rgb or vec(x,y,z) return value should offer plenty of possibilities :)
:beer:

Title: Re: Spherical Texture Fold
Post by: Aexion on July 13, 2011, 12:12:54 PM

Just some simple ideas for this:
1. For low resolution maps, interpolations (unless you want large square blocks..)
2. The surface color can come from the last iteration texture sampling
3. Put graffiti on your fractal walls..
4. If the texture (or bump map) color is 0 break the iteration loop early, so it becomes a hole..
5. Do not put banner ads on your fractal walls ("if you like this fractal, buy etc,etc,etc...")..

:)

Title: Re: Spherical Texture Fold
Post by: DarkBeam on July 13, 2011, 07:58:39 PM

Ahhh another crazy striking idea! :D
You are incredible :D
I am developin new formulas but internet is hard to use. hopefully cqn upload stuff soon with my pocketpc! :(

Title: Re: Spherical Texture Fold
Post by: Madman on July 13, 2011, 07:59:49 PM

That would be sooooooooh nice. Can't wait for the next M3D update Jesse ;D No pressure intended :dink:

Title: Re: Spherical Texture Fold
Post by: Jesse on July 13, 2011, 10:38:35 PM

Quote from: Aexion on July 13, 2011, 12:12:54 PM

We might regret this option when reading silly stuff everywhere... but we know who to blame :dink:

To 4: for a hole you normally have to bailout in escapetime formulas, so it would become solid, if this works...

Quote from: DarkBeam on July 13, 2011, 07:58:39 PM

Ahhh another crazy striking idea! :D
You are incredible :D
I am developin new formulas but internet is hard to use. hopefully cqn upload stuff soon with my pocketpc! :(

Don't wanted to bother you in your holidays, but since i get you here a wish:
In Box modifications with a power function, could you skip the power calculation on power=1?
This would speedup very much and you could simply copy the code from my ABoxVaryScale formula.
:beer: :beer: :beer:

And a question to all:
Does anybody know where i can get some fullsphere lightmaps in spherical projection for using it in the program for free?
Would be nice to have some basic ones...

Title: Re: Spherical Texture Fold
Post by: DarkBeam on July 13, 2011, 11:02:07 PM

In future releases rpow is replaced with inversion power, a big speedup and more variations :)
Also I fixed a "error" in your planefold and made a sort of origami fold formula, will see soon i hope :D

Title: Re: Spherical Texture Fold
Post by: Jesse on July 15, 2011, 10:07:35 PM

I encountered a problem with maps and the 4 point DE method:
It seems that the plenty of pixeledges affects the DE strongly, that even a very low stepwidth multiplier can't help.
Without double or more checking and giving the bigger gradients a chance, i fear that mapping works only with analytical DE's.

Attached is a pic with a map translation function and then just bailing out with a power 3 on the veclength function (marble).
The noise can't be really avoided here.

Quote from: DarkBeam on July 13, 2011, 11:02:07 PM

In future releases rpow is replaced with inversion power, a big speedup and more variations :)

This sounds interesting, i like to see this. :)

Welcome to Fractal Forums

Fractal Software => 3D Fractal Generation => Topic started by: Aexion on July 12, 2011, 01:47:13 PM