Runevision on Procedural Generation

I’ve linked to Rune Skovbo Johansen’s blog before. This time rather than a tutorial, it’s a look at three approaches to procedural generation: simulation, functional, and planning. 

It’s useful to consider the fundamental differences between algorithms. There’s no one best way to do procedural generation, and knowing the tradeoffs between your various options is an important part of making the best choices when you implement your procedural generation.

I think its important that the post highlights that the simulation approach doesn’t require you to simulate the exact thing you want for your result. Maybe your plate tectonic simulator can be re-purposed to also generate the fossa on Mars, or your cellular automaton can place the clouds over your city or the birds in the park.

Likewise, the functional approach has some really useful properties that comes from each sampled point being independent from the other points. This means that is can be calculated in real time and in parallel. The blog post also goes into a bunch of other considerations for choosing between the three approaches. Do go read it.

Can you think of examples of these different categories? What about approaches to procedural generation that don’t fall into these three categories? I’d love to hear any examples you can think of.

http://blog.runevision.com/2015/08/procedural-world-potentials-simulation.html