0

I recently implemented Ian Parberry's Exponentially Distributed Terrain algorithm in Javascript. I've checked it against his c++ code, but when generating multiple octaves I get only circular hills and valleys on smaller at scales and intervals, and no gradual features like in the screenshots provided in the paper.

In theory, this makes sense, since adding the exponentially distributed coefficient only decreases the frequency of large gradients, which makes geographical features more distinct at every level, but I'm unsure how the provided screenshots were generated except by combining the algorithm with something like Perlin Noise or Midpoint Displacement.

In short, does anyone who's implemented this know what I'm talking about? Or must there be a bug in my code?

Link to paper: http://jcgt.org/published/0004/02/01/

Screenshots:

Terrain generated with 12 octaves Terrain generated with 12 octaves

Terrain generated with 4 octaves Terrain generated with 4 octaves

Perlin sample for reference My Perlin terrain for reference

Combined perlin and exponentially distributed Combined perlin and exponentially distributed

Elias N-d
  • 199
  • 1
  • 1
  • 4
  • 2
    Unless someone has had your specific issue before it seems unlikely anyone will be able to help you check your implementation of the algorithm without links or examples of the source. Even without knowledge of the algorithm developers may spot logical errors or flaws in the code / math (such as rounding issues). – Michael Whinfrey Aug 20 '19 at 23:38
  • Yeah, I figured -- I'm looking more for confirmation that this is in fact a problem and not intended behaviour before I go back to scouring my code for a bug. – Elias N-d Aug 21 '19 at 00:04

0 Answers0