Hi,

I've been trying out different noise/random functions. I started to use noiseDetail, but something happened that I can't explain and after searching inside the code it turned out it was the falloff, which remains changed onces changed. See the code example below.

Processing 1.09.

Please tell me if I am doing something wrong or if this is a bug. Thanks.

Jaden


Code:

void setup() {
  size(100,100);

  // noiseDetail falloff set to 0
  // output ranges between 0 and 1
  for (int i=0; i<3;i++) {
    noiseDetail(4,0);
    float x = noise(i);
    println(x);
  }

  // noiseDetail falloff set to 25
  // output ranges in the 100s and 1000s
  for (int i=0; i<3;i++) {
    noiseDetail(4,25);
    float x = noise(i);
    println(x);
  }

  // noiseDetail falloff set to 0 again
  // but output STILL ranges in the 100s and 1000s!
  for (int i=0; i<3;i++) {
    noiseDetail(4,0);
    float x = noise(i);
    println(x);
  }

}

void draw() {
  exit();
}
 

...am only confirming/restating philo's reply...

Your first block, with falloff==0, fails to set the falloff, because it's zero.
Your second block, does set the falloff to 25 (an unusual, but allowable, value)
Your third block again fails to set the falloff,  because it's zero.

Recode your blocks using (for example) the values 0.1,0.75,0.3 and you should note a difference in each block.

A falloff of zero means subsequent octaves would give NO additional contribution, so it's a "nonsense" value -- it is equivalent to calling noiseDetail(1) *without* concern for the falloff value.

OTOH, falloff values are *not* checked to be <= 1.0 (though the reference might suggest they are).  But values > 1.0 produce higher amplitudes in the higher octaves, which is the opposite of how most people expect a Perlin noise function to behave.  (such noise will be very 'noisy', less smooth that you might expect)

Further, falloffs >0.5 will return values outside the 0..1 range - that's why your values are in the 100's and 1000's using a falloff of 25.  If you DO wish to "scale" the returned values into the 100's, you should instead do something like:
float x = noise(i) * 100.0;

hth