Procedural Generation, Intro

This is the first of what I hope to be many posts exploring the topic of procedural generation, particularly as it applies to game development and art.

At it simplest level, procedural generation (pg) is the use of a small amount of code, or an algorithm, to create a result, rather than creating that result by hand. Randomness and pseudo-randomness generally figure into the process, as well as set theory, emergence, and a wide variety of mathematical concepts such as fractals, the Fibonacci sequence, cellular automata, Perlin noise algorithms, and occasionally cryptography.

PG starts with the creation of a series of bits or numbers, then branches out into the myriad uses to which that series can be applied. How the numbers are chosen is just as important. So PG starts a level lower, at the algorithm which creates the data.

A list of numbers can mean almost anything depending on its context. But for a given context, not all sets of numbers will work. Therefore it is important to have a number generator which will produce useful data for a given task. This is where experimentation comes in to play.

But enough of the high-level stuff.

I have several years of notes, graphics, experiments, and source code through which I am currently sorting. Over the upcoming months I will post breakdowns of some of them, particularly those which can be applied to game development. And in those, I will be providing ideas about how to make PG useful, and how to tweak things so that using this method actually saves time and effort. Here are some of the ideas which I will cover:

  • terrain generation
  • town placement
  • resource placement
  • maze generation
  • cave/dungeon generation and population
  • place name generation
  • graphics creation
  • plant/tree generation

…and various combinations of the above.

In the meantime, click here to see the nearly 30 old entries I have made in this blog regarding procedural generation.