Loading...
Logo
Processing Forum
Contributed Library Questions
jonobr1
jonobr1

[Toxiclibs] PerlinNoise to Specified Target Destination

in Contributed Library Questions  •  2 years ago  
This particular example is using toxiclibs, but could just as easily be done using PVector, so I wasn't sure whether to place this in Programming Questions or Contributed Library Questions.

Anyways, onto the question:

Below is code to generate a path from (0, height) to an unknown destination. I was trying to use the highlighted line to steer the path towards the target Vector2D  t, but it doesn't prove to be reliable.

Copy code
  1. import toxi.geom.Vec2D;
  2. import toxi.math.noise.PerlinNoise;

  4. PerlinNoise n;
  5. Vec2D   p, pp, t;
  6. float   noiseScale = 300,
  7.         noiseStrength = 10,
  8.         stepSize = 5,
  9.         angle;
  10. boolean atTarget = false;

  12. void setup() {
  13.   size(800, 600);
  14.   smooth();
  15.   setPerlinNoise();
  16. }

  18. void draw() {
  19.   
  20.   stroke(0);
  21.   fill(255);
  22.   if(!atTarget) {
  23.     pp.set(p);
  24.     
  25.     angle = n.noise(p.x/noiseScale, p.y/noiseScale) * noiseStrength;
  26.     
  27.     // Create our FlowField position
  28.     p.addSelf(cos(angle) * stepSize * 3,
  29.               sin(angle) * stepSize * 3);
  30.     
  31.     // Steer towards our destination
  32.     p = p.interpolateToSelf(t, .1);
  33.     
  34.     // Draw where we've been
  35.     ellipse(p.x, p.y, 5, 5);
  36.     line(pp.x, pp.y, p.x, p.y);
  37.     
  38.     // Check to see if we're within target
  39.     if(abs(t.x - p.x) < 25 && abs(t.y - p.y) < 25) {
  40.       atTarget = true;
  41.     }
  42.   }
  43. }

  45. // Click to reset
  46. void mouseReleased() {
  47.   setPerlinNoise();
  48. }

  50. void setPerlinNoise() {
  51.   
  52.   background(255);
  53.   
  54.   // Reset variables
  55.   p  = new Vec2D(0, height);
  56.   pp = new Vec2D(p.x, p.y);
  57.   t  = new Vec2D(width/2, height/2);
  58.   n  = new PerlinNoise();
  59.   
  60.   atTarget = false;
  61.   
  62.   p.set(0, height);
  63.   n = new PerlinNoise();
  64.   
  65.   // Draw our target region
  66.   noStroke();
  67.   fill(255, 200, 0);
  68.   ellipse(t.x, t.y, 50, 50);
  69. }
Is there a way to get the path to hit the target 100% of the time? Perhaps I don't fully understand  interpolateToSelf() and   interpolateTo() . Well more precisely, I don't fully understand the difference between those two methods. Maybe the problem lies there?

—
http://jonobr1.com/

Replies(3)

nardove
nardove

Re: [Toxiclibs] PerlinNoise to Specified Target Destination

2 years ago

What the interpolate methods do is to gradually move your vector to the target vector by a factor of 0 < n < 1 on every step, so for example you specify the target vector, then you specify the factor which will dictate how many steps will take to get to the destination, if you choose 1.0 the vector will jump to the target in 1 step along the way but if you select 0.5 the vector will move to half the distance to its target, them will move half of the remaining distance and the process continue until it reach its target, 

Run the following code and play with the factor value (line 33) to get the idea

Copy code
  1. import toxi.geom.Vec2D;
  2. import toxi.math.noise.PerlinNoise;

  4. PerlinNoise n;
  5. Vec2D   p, pp, t;
  6. float   noiseScale = 300,
  7.         noiseStrength = 10,
  8.         stepSize = 15,
  9.         angle;
  10. boolean atTarget = false;

  12. void setup() {
  13.   size(320, 240);
  14.   smooth();
  15.   setPerlinNoise();
  16. }

  18. void draw() {
  19.   
  20.   stroke(0);
  21.   fill(255);
  22.   if(!atTarget) {
  23.     pp.set(p);
  24.     
  25.     /*
  26.     angle = n.noise(p.x/noiseScale, p.y/noiseScale) * noiseStrength;
  27.     
  28.     // Create our FlowField position
  29.     p.addSelf(cos(angle) * stepSize, sin(angle) * stepSize);
  30.     */
  31.     
  32.     // Steer towards our destination
  33.     p = p.interpolateTo(t, 0.5);
  34.     
  35.     // Draw where we've been
  36.     line(pp.x, pp.y, p.x, p.y);
  37.     ellipse(p.x, p.y, 5, 5);
  38.     
  39.     
  40.     // Check to see if we're within target
  41.     if(dist(p.x, p.y, t.x, t.y) < 25) {
  42.       atTarget = true;
  43.       println( "at target" );
  44.     }
  45.   }
  46. }

  48. // Click to reset
  49. void mouseReleased() {
  50.   setPerlinNoise();
  51. }

  53. void setPerlinNoise() {
  54.   
  55.   background(255);
  56.   
  57.   // Reset variables
  58.   p  = new Vec2D(0, height);
  59.   pp = new Vec2D(p);
  60.   t  = new Vec2D(random(width/2, width), random(0, height));
  61.   n  = new PerlinNoise();
  62.   
  63.   atTarget = false;
  64.   
  65.   p.set(0, height);
  66.   n = new PerlinNoise();
  67.   
  68.   // Draw our target region
  69.   noStroke();
  70.   fill(255, 200, 0);
  71.   ellipse(t.x, t.y, 50, 50);
  72. }

Now the problem we have here is that the angle is distorting the p vector too much and it will never reach the target, the times it does is just by luck, Im very interested in solving this problem, so I will post any progress when I get a chance

Cheers

- rS

toxmeister
toxmeister

Re: [Toxiclibs] PerlinNoise to Specified Target Destination

2 years ago
Your example is relying on 2 separate forces which need to be combined in a balanced way. In your case you manipulated the position directly and once you get closer to the target, the target attraction is simply not strong enough anymore to have any impact against the noise() based counter force. 

For example: You used interpolateToSelf() with a 0.1 value meaning you approach the target pos only by 10% of the current distance. As your distance shrinks so does the change amount. On the other hand your noise dir always moves as a fixed speed regardless of distance to the target -> BAD combination! :)

Below is an updated version of the draw() function showing a better/correct way:
Copy code
  1. // udpated these 2 params
  2. float noiseScale = 10;
  3. float noiseStrength = TWO_PI;

  5. void draw() {
  6.   stroke(0);
  7.   fill(255);
  8.   if(!atTarget) {
  9.     pp.set(p);
  10.     
  11.     // compute flow direction based on pos in noise space
  12.     angle = n.noise(p.x*noiseScale, p.y*noiseScale) * noiseStrength;
  13.     Vec2D flowDir=Vec2D.fromTheta(angle);

  15.     // and direction to target
  16.     Vec2D targetDir=t.sub(p).normalize();

  18.     // now combine forces in a 50:50 ratio
  19.     // reducing the interpolation factor below 0.5 will NOT guarantee
  20.     // that the target is ever being reached
  21.     // any value >0.5 means the target attraction is stronger than the flow dir
  22.     Vec2D dir=flowDir.interpolateTo(targetDir,0.5);

  24.     // normalize to step size and update position
  25.     p.addSelf(dir.normalizeTo(stepSize * 3));
  26.     
  27.     // Draw where we've been
  28.     ellipse(p.x, p.y, 5, 5);
  29.     line(pp.x, pp.y, p.x, p.y);
  30.     
  31.     // Check to see if we're within target
  32.     if(p.distanceTo(t) < 25) {
  33.       atTarget = true;
  34.     }
  35.   }
  36. }

jonobr1
jonobr1

Re: Re: [Toxiclibs] PerlinNoise to Specified Target Destination

2 years ago
This makes so much more sense when thinking about ratios of directions! I was thinking that because interpolateTo was following the noise it would always move 10% closer to the target effectively bringing the p closer to t every frame. Thanks very much, both @nardove and @toximeister, for the explanations.

@toximeister, is there a way to get the source of toxiclibs other than the mercurial repo?

—
http://jonobr1.com/
Top Reply

Statistics

  • 3 Replies
  • 2043  Views
  • 2 Followers

Actions

Related Posts