Procedural Island Generation (iii)

People can try something fully physics-based (or rather, physics-inspired) even for earlier stages, but there are problems:

* You still need some kind of nondeterministic input so you don't always generate the same world.

* You must do the whole world at once, rather than being able to generate each area independently.

* This requires the computation to run for a long time, and needs to feed back in on itself (think of "lake overflows a natural dam and carves a valley, then the tectonics lift it and change the low point anyway").

* It's very easy for your code to result in "boring" outputs, such as "all flat" or "infinitely deep valleys".

The result is incredibly detailed terrain that is completely unique based on the initial parameters and the randomness of the time-elapsed process that is non-deterministic.

The only downside is that it takes hours to simulate and that's more than most folks are interested in investing in.

Here are some examples of this kind of software:

World machine: https://www.world-machine.com/

Terragen: https://planetside.co.uk/

Vue: https://www.bentley.com/software/e-on-software-free-download...

Gaia: https://www.procedural-worlds.com/products/indie/gaia/

It'd be neat to see a game world where the simulation remains ongoing, where the world is actively changing.

https://www.google.com/search?q=news.ycombinator.com+procedu...

This one is a particularly useful starting point: https://news.ycombinator.com/item?id=5196154

or infuriating because some small detail is not right causing all of the biomes to die off.

Sometimes that's what people want, and sometimes they want something they can control more strictly.

There's a different blog series that does go into plate tectonics: https://frozenfractal.com/blog/2023/11/13/around-the-world-2...

Coincidentally Civ 7 just announced they're using plates with voronoi for their new map generator.

I'm not sure what you mean here. A lot (most?) procedural systems use Voronoi methods for procedural terrain generation, so this isn't some wholly novel solution. From what I can tell (haven't looked into it super deeply), it looks like Lloyd's algorithm is basically a variation of random bisections.

The choices the author made reflect the way they conceptualize the generating process. This is in contrast to a data-informed approach.

A data-informed approach to island generation would be a spatially auto-correlated generating model whose parameters were derived from real life data on islands. You would then be able to generate more in-distribution islands that were statistically similar to islands in the corpus.

You don't often see this in the proc-gen world because part of the fun is adding epicycles and various tunable knobs.