Poisson Disk Sampling?

I was thinking that Poisson disk sampling might be a nice Code device project to start with. Before doing so, I thought I should ask if the Probability device uses it. That’s unclear in the Home | Device Reference | Filter Devices | Probability reference.

Probability

The Probability filter treats the input map as a probability density function, and creates an output mask that randomly scatters single pixel dots according to that density function.

This can be useful for object placement maps as well as texturing purposes.

Device Parameters

  • Probability Type
    What kind of distribution model to use. There is currently only one option.
    • Uniform – Use a uniform distribution to scatter the points across the map.
  • Bias
    Bias adjusts what value in the input map has a 50% chance to create a pixel instance. Adjusting this will cause greater or fewer numbers of dots to be created. A bias of 0 will produce no output, and a bias of 1 will produce an all-white mask. The default is 0.5.
  • Strength
    After the bias is applied, further reduce the probability that a dot is created. A value of 1.0 has no effect, while a value of 0 will ensure that no dot is created.

Blue noise is very useful!..

The probability device currently only uses white noise. There is, in fact, an in-progress experimental device that implements it. But don’t let that stop you!

2 Likes

Blue noise is also rather expensive, too, isn’t it? I will at least give it a try!

2 Likes