Glitch in Turing pattern fragment shader
in
Programming Questions
•
7 months ago
I was trying to write a fragment shader for the quick generation of multi-scale fractal turing patterns, as described in Jonathan McCabes paper
here and all those wonderful sketches at openprocessing. For some reason, when I try to run th ketch, the entire computer freezes? I can't really figure it out, and I'm not getting any error messages. I feel like I'm probably just missing something simple in the shader, but would you mind looking at the code?
The Processing portion:
Copy code
- float[] aR= {
- 50.0, 5.0,2.0
- };
- float[] iR= {
- 100.0, 10.0,4.0
- };
- float[] small= {
- .4, .2,.1
- };
- PShader s;
- float i;
- void setup() {
- size(640, 360,P2D);
- s = loadShader("turing.glsl");
- s.set("activatorRadii",aR);
- s.set("inhibitorRadii",iR);
- s.set("smallAmount",small);
- i=1;
- loadPixels();
- randomize();
- }
- void randomize(){
- for(int i=0;i<pixels.length;i++)
- pixels[i]=color(random(0,255));
- updatePixels();
- }
- void draw() {
- s.set("iterations",i);
- filter(s);
- i++;
- }
Copy code
- #ifdef GL_ES
- precision mediump float;
- precision mediump int;
- #endif
- #define LEVELS 3
- uniform sampler2D textureSampler;
- uniform vec2 texcoordOffset;
- uniform float activatorRadii[LEVELS];
- uniform float inhibitorRadii[LEVELS];
- uniform float smallAmount[LEVELS];
- uniform float iterations;
- varying vec4 vertColor;
- varying vec4 vertTexcoord;
- float grid(vec4 v){ //maps the red value of the texture to between -1 and 1
- return v.r*2.0-1.0;
- }
- vec4 unGrid(float g){ //maps the new value (which is most of the time between -1.4 and 1.4) to between -1 and 1
- return vec4(vec3(g+1.4)/2.8,1.0);
- }
- vec4 tex(float x, float y){ //simply gets the color of the texture at (x,y).
- return texture2D(textureSampler,vec2(x,y));
- }
- int constrained(float x, float y){ //makes checks to make certain that it is only using pixels on the texture
- if(x>=0.0&&x<=1.0&&y>=0.0&&y<=1.0)
- return 1;
- return 0;
- }
- void main(void) {
- vec4 pixelColor = texture2D(textureSampler,vertTexcoord.xy);
- vec2 c=vertTexcoord.xy;
- vec2 d=texcoordOffset;
- float activator[LEVELS];
- float inhibitor[LEVELS];
- float variations[LEVELS];
- int bestVariation=0;
- for(int i=0;i<LEVELS;i++){
- activator[i]=0.0;
- inhibitor[i]=0.0;
- float count=0.0;
- //these lines make activator[i] equal to the average value of the pixels in a radius of activatorRadii[i]
- for(float x=c.x-activatorRadii[i]*d.x;x<c.x+activatorRadii[i]*d.x;x+=d.x){
- for(float y=c.y-activatorRadii[i]*d.y;y<c.y+activatorRadii[i]*d.y;y+=d.y){
- if(constrained(x,y)==0){
- activator[i]+=grid(tex(x,y));
- count+=1.0;
- }
- }
- }
- activator[i]/=count;
- count=0.0;
//these lines make inhibito[i] equal to the average value of the pixels in a radius of inhibitorRadii[i]- for(float x=c.x-inhibitorRadii[i]*d.x;x<c.x+inhibitorRadii[i]*d.x;x+=d.x){
- for(float y=c.y-inhibitorRadii[i]*d.y;y<c.y+inhibitorRadii[i]*d.y;y+=d.y){
- if(constrained(x,y)==0){
- inhibitor[i]+=grid(tex(x,y));
- count+=1.0;
- }
- }
- }
- inhibitor[i]/=count;
- //these lines seek out the level of the turing pattern with the smallest difference between the activator and inhibitor
- variations[i]=abs(activator[i]-inhibitor[i]);
- if(variations[i]<variations[bestVariation])
- bestVariation=i;
- }
- float g=grid(tex(c.x,c.y));
- if(activator[bestVariation]>inhibitor[bestVariation])
- g+=smallAmount[bestVariation]/iterations;//dividing by iterations ensures that as time goes on, there will be less impact. This is how I adjusted to the difficulty of mapping the grid between -1 and 1 each time.
- else
- g-=smallAmount[bestVariation]/iterations;
- gl_FragColor = unGrid(g);
- }
Also, if this has any impact, I'm using Processing 2.0b6 on a mac os x 10.8.2 with a NVIDIA GeForce 320M graphics card.
Any help would be wonderful!
1
