Recently, I went down a rabbit hole attempting to learn how to generate interesting looking maps for games (insert mind-blown-gif here). While I’m not going to be using Voronoi diagrams anytime soon, I am still interested in attempting to generate Civilization-like maps.
Most map generation algorithms rely on perlin noise. I wanted to go a different route.
So far, these are the rules that I’ve come up with for my algorithm:
1. Create a grid of some given size (e.g., 100 x 50). Set all tiles to “ocean”.
2. Randomly pick 8 tiles (variable) across the map to seed as “land”.
3. Now iterate across all tiles on the map and build an array of tiles that are “ocean” tiles but have a land tile N/W/S/E of them.
4. Once you have this array, randomly pick a single value. Flip it from “ocean” to “land”
5. Repeat step 3 until a given number of tiles have been filled across the grid (e.g., for an “Earth-like planet”, 30% of tiles will be land).
6. Now, to add some additional randomness — iterate across all tiles and build an array containing all “land” tiles next to water (N/W/S/E).
7. Randomly pick 1, flip from land to ocean.
8. Repeat step 6 a given number of tries (e.g., 100).
9. Done! (Maybe)
Height maps, biomes and all that can come later. To quote Amit, whom I linked to earlier:
The most realistic approach would have been to define elevation first, and then define the coastline to be where the elevation reaches sea level. Instead, I’m starting with the goal, which is a good coastline, and working backwards from there.
Anyway, it’s been kind of neat to figure out.
Another example map, created using the above rules:
