Generate hexagonal islands

Question

I'd like to generate flat islands existing of multiple hexagons. So far I've been able to create hexagonal meshes with code, but can't figure out how to position them in groups, creating a randomly shaped island. (Any shape that's not a perfect circle, square etc.) I think I would need an algorithm, that places hexagon tiles next to multiple sides of an existing tile. If you can help me with an idea for an algorithm, then that would be great.

Answer 1

Are you looking for something like this?

Place 1 hexagon.
for i in (islandSize-1):
    Scan all hexagons for open sides. Place open sides in a list named hexBorders

    Choose a random index in hexBorders, attach a new hexagon there

That algorithm should give you a fairly roundish island, roughly centered on the original hex, because older hexes have more chances to get picked.

You can tune this shape by preferring either newer or older hexagons (e.g. you could include hexagon age in hexBorders, and adjust your random choice so it prefers younger hexes).

Answer 2

Recently I was also doing random map generator for tile based map and hit a wall while try to add more advanced features (in tile space) the realism of output was not good. I decided to create a 2D/3D image based map and then convert it to tile map. Still not finished with adding all the features I want but the result is already a magnitude better then before:

1. map generator

   see my simple random map generator in C++. It is based on Diamond&Square algorithm with some tweaking to obtain island like maps.

2. conversion to tile-map

   Simply map Cartesian pixel into your hexagonal grid layout. You can also compute the average of some area instead of using single pixel per cell/tile.

   For 3D tile maps this will produce "voxel-ated" output so you need to add additional filtering see

   • How to procedurally generate tile map for some ideas.

Answer 3

Since is a pretty open ended question, this article by Red Blob Games about hexagonal data structures would be an excellent place to start. The author describes how you can use 2D arrays to store the hexagons, and how you can iterate through them.

Once you understand the relation of hexagons to one another you can start to iterate through them in interesting ways.

Probably the easiest way to generate an "island" would be with a SIR-type model, also known as an epidemic model. This is a model that is commonly used by researchers to simulate the spread of infectious disease, but I've found that you can also use it to generate pseudo-natural shapes (like an island!). SIR stands for Susceptible-Infectious-Recovered. Those are the three states of a "cell", or in this case hexagon. At any given step of the algorithm, an infected cell can infect a neighboring cell. Think about it like this: at the start of your algorithm, one hexagon is "infected" (land) and the rest are not (water). At each iteration of the algorithm, cells adjacent to an infected cell have a chance (say, 1 in 10) of being infected as well (turning into land). After many iterations, you'll find that the shape of the infected group of hexagons is pretty random looking, but they're all touching. For a grid-bsed example, here's some images I've uploaded to imgur. Pseudo-code for this algorithm is below.

cellsToDo = [originCell]
for 100 iterations:
  for each cell in cellsToDo:
    for each neighbor to the current cell:
      if randomValueBetween(0, 10) == 1:
        set the current cell as infected
        add the current cell to the cellsToDo list

There are definitely other algorithms, but I'd start with learning how the hexagons are related to each other and can be stored.