I got the new form of chunk generation working in my terrain-test branch. I’m probably going to bring this over into the main branch, but what I wanted to work on was the actual terrain generation. The first chunk generated (hexagon coordinates q=0, r=0, s=0) for it’s “center” or anchor will always have the starting city core at it, but I also want other city cores spread across the map. At what distances apart they spawn will need to be determined experimentally by playing, and if I add difficulty settings, they could affect this distance calculation.
The first thing I tackled was enemy core generation, and my first pass at an algorithm showed the I need to make things more random. I’m using a hash function with a seed appended to it, but the results were a little too deterministic for my liking.

This is exactly why I wanted to create something to give me a larger view of the terrain map, because I might have missed this regularity otherwise.
I’m using Python’s hash function on a tuple containing the hexagon coordinates (q, r, s + seed) and then checking if a hex should maybe have a core added to it if the hash is % some number. Obviously this isn’t working yet, but it’s a fine first pass.
I also started working on terrain hex generation. Ideally I’d like blue hexes (water) to cluster together a bit to give water bodies, and the other colours to have some other nice distribution. The first pass uses OpenSimplex to generate the noise, due to Perlin noise having some artifacts, and then binning the values directly to sprites. I also spent too much time reading about procedural terrain generation and not programming.

Again, the first pass here was less of what I was looking for, but it at least works and I can tweak the normalization function until I’ve got something better to look at with the correct proportions of water hexes. I do plan on having water be a barrier of movement to certain units, and stop building of normal networks.