Terrain/Mountain algorithm not working as intended

Question

I want to create a terrain with a mountain on it, using a very basic principle, shown by this height mapping:

0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 1 0 0 0 0 0
0 0 0 0 1 2 1 0 0 0 0
0 0 0 1 2 3 2 1 0 0 0
0 0 1 2 3 4 3 2 1 0 0
0 0 0 1 2 3 2 1 0 0 0
0 0 0 0 1 2 1 0 0 0 0
0 0 0 0 0 1 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0

It starts at a random point with height = 4, and then gradually decreases it amonst the neighbours.

The recursive idea is simple, I start a point, recurse to the top/down/left/right with height - 1 (in this example), and only if not encountered yet, I set their values.

I implemented it as follows:

private void createMountain(final float[][] heightMapping, final float startHeight) {
    boolean[][] traversed = new boolean[width][depth];
    boolean positive = (startHeight >= 0f);
    int x = random.nextInt(width);
    int z = random.nextInt(depth);
    recursiveUpdate(heightMapping, traversed, x, z, startHeight, positive);
}

private void recursiveUpdate(final float[][] heightMapping, final boolean[][] traversed, final int x, final int z, final float startHeight, final boolean positive) {
    if (x < 0 || x >= width || z < 0 || z >= depth) {
        return;
    }
    if (traversed[x][z]) {
        return;
    }
    if ((positive && startHeight <= 0f) || (!positive && startHeight >= 0f)) {
        heightMapping[x][z] = 0f;
        return;
    }
    traversed[x][z] = true;
    heightMapping[x][z] = startHeight;
    recursiveUpdate(heightMapping, traversed, x, z - 1, calculateNewStartHeight(startHeight, positive), positive);
    recursiveUpdate(heightMapping, traversed, x, z + 1, calculateNewStartHeight(startHeight, positive), positive);
    recursiveUpdate(heightMapping, traversed, x - 1, z, calculateNewStartHeight(startHeight, positive), positive);
    recursiveUpdate(heightMapping, traversed, x + 1, z, calculateNewStartHeight(startHeight, positive), positive);
}

private float calculateNewStartHeight(final float startHeight, final boolean positive) {
    float delta = startHeight / maxDecayFactor;
    return (positive) ? startHeight - delta : startHeight + delta;
}

However it gives me the following output:

0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  
0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  
0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  
0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  
0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  
0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  
0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  
0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  
0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  
0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  
0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  
0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  
0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  
0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.1  
1.9  1.6  1.2  1.0  0.8  0.6  0.5  0.4  0.3  0.3  0.2  0.2  0.1  0.1  0.1  0.1  
2.4  3.0  3.8  4.7  5.9  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.0  0.1 

With the issue being that it forms a line now, which was not my intention, instead of gradually smoothing out.

How can I implement my desired algorithm?

Answer 1

The problem with your recursive approach that it is basically performs a DFS, so you always go in one direction, following the longest branch. But this branch is always decaying.
Since you are also maintaining a traversed set - it falsely ensures that you don't visit the same vertex later on for the other branch (the other recursive call).

There are 2 ways to solve this issue:

1. More elegant and probably more efficient - change your algorithm from a DFS to BFS. change cells at distance 1 from the source, then cells at distance 2, and so on...
2. Less elegant - but will require minimal changes to the code: change the stop condition of your algorithm, instead of if (traversed[x][z]) { return; } do something like if (heightMapping[x][z] > startHeight) { return; }. This will ensure you can update the height if it should be higher and work as intended.

---

The BFS updating should be something like (pseudo-code):

Q <- new Queue() //or even better - priority queue that holds the highest point at the top
Q.push((x,y,height)
visited[width][depth]; //init as all false
while Q.empty() == false:
   curr <- Q.pop()
   if (sanity check for x<0 , y< 0 ,..):
      continue
   if visited[x][y] == true:
      continue
   if height <= 0: //or some epsilon instead of 0 if there are floating point issues
      continue
   heights[x][y] = height
   visited[x][y] = true
   Q.push(x+1,y,calculateNewHeight(...))
   ... //similarly for all directions

Answer 2

There's no need for recursion.

Say you want the top of your mountain to be height h at position x, y.

height(x,y) = h
for dist = 1 to h-1
  for x' = x - dist to x + dist
     x_dist = abs(x - x')
     y_dist = dist - x_dist
     height(x', y + y_dist) = h - dist
     height(x', y - y_dist) = h - dist

Your height at any given distance from the peak is the height of the mountain less the orthogonal distance from the peak. I'm assuming you're starting with all zeroes so those don't need to be set if you're sufficiently far from the peak.