I'm trying to work out how to do so my player position takes into the account of the floor so if I generate a floor that is hilly when I walk I won't walk through the heel. The problem is that I'm not sure how to do it.

I can do basic collision detection for walls. However, this isn't collision detection. I need the add the height of the terrain to the player.

So as you see the terrain is really a grid with 4 points making a face. So I can get the 4 points and work out. I can probably work this out on myself, but I was wondering if I should store say the heigh at (x,z) in a 2d array or do I just work out the height again everytime I move?

It just I use simplexnoise and I worry that this will be expensive to do every turn.

 import java.util.ArrayList;

    public class VoxelLevel 
    Mesh mesh;
    Material material;
    Transform transform;
    Shader shader;

    VoxelLevel(String textureName)
        material = new Material(new Texture(textureName));
        transform = new Transform();

        shader = BasicShader.getInstance();



    private void generateLevel() 
        ArrayList<Vertex> vertices = new ArrayList<Vertex>();
        ArrayList<Integer> indices = new ArrayList<Integer>();

        int sizeX = 10;
        int sizeZ = 10;
        int freq = 25;

        for(int x=-300; x<300; x+= sizeX)
            for(int z=-300; z<300; z+=sizeZ)

                vertices.add(new Vertex(new Vector3f(x,50+25*(float)(SimplexNoise.noise(x/freq, z/freq) + 0.25 * SimplexNoise.noise(2*x/freq, 2*z/freq)),z), new Vector2f(0.75f,0.75f)));
                vertices.add(new Vertex(new Vector3f(x+sizeX,50+25*(float)(SimplexNoise.noise((x+sizeX)/freq, z/freq) + 0.25 * SimplexNoise.noise(2*(x+sizeX)/freq, 2*z/freq)),z), new Vector2f(0.75f,1.0f)));
                vertices.add(new Vertex(new Vector3f(x,50+25*(float)(SimplexNoise.noise(x/freq, (z+sizeZ)/freq)+ 0.25 * SimplexNoise.noise(2*x/freq, 2*(z+sizeZ)/freq)), z+sizeZ), new Vector2f(1.0f,0.75f)));
                vertices.add(new Vertex(new Vector3f(x+sizeX,50+25*(float)(SimplexNoise.noise((x+sizeX)/freq, (z+sizeZ)/freq)+ 0.25 * SimplexNoise.noise(2*(x+sizeX)/freq, 2*(z+sizeZ)/freq)),z+sizeZ), new Vector2f(1.0f,1.0f)));

        Vertex[] vertArray = new Vertex[vertices.size()];
        Integer[] intArray = new Integer[indices.size()];


        mesh = new Mesh(vertArray, Util.toIntArray(intArray));

    public void input()


    public void update()
        //SimplexNoise.noise(0.25, 0.25);
        //SimplexNoise.noise(0.35, 0.65);
        //SimplexNoise.noise(0.254, 0.225);
        //SimplexNoise.noise(0.254, 0.215);

    public void render()
        shader.updateUniforms(transform.getTransformation(), transform.getProjectedTransformation(), material);

  • \$\begingroup\$ Sounds like you should try both ways and see what works for you. The only problem I see you describing here is a design indecision problem. And that's a problem you need to solve on your own. \$\endgroup\$
    – House
    Commented Oct 8, 2013 at 0:21
  • \$\begingroup\$ Byte56 I was more worried about computational costs. Like I have to call PerlinNoise 4 times every frame versus say storing a 100x100 array or even 4 * 100x100 array. I was under the impression that PerlinNoise was expensive calculation. But, it seems to have no effect on my fps. But, then I was told that fps is misleading. \$\endgroup\$
    – pangaea
    Commented Oct 8, 2013 at 0:40
  • \$\begingroup\$ Profiling your code is a good way to find out if one way is faster than another. It's more accurate than an answer you'll get here since it's actual instead of theoretical. Netbeans has a nice profiler. \$\endgroup\$
    – House
    Commented Oct 8, 2013 at 1:39
  • \$\begingroup\$ I don't think you need to profile to know that a array look up is going to be faster then calculating the perlin noise at a given point. Store the vertArray and get the points from that. Something like i0 = (x * 600 + z) * 4 \$\endgroup\$
    – nykwil
    Commented Oct 9, 2013 at 20:45

1 Answer 1


Getting the height from your terrain at any x/z point is super fast so there's no reason to store and manage a cache, it just adds complexity.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .