Lately I’ve been following the work of three people who are experimenting with developing game-worlds procedurally. That is, they’re getting the computer to generate the worlds with minimal input from them, as opposed to the very manual way most game-worlds are produced.
This way of producing virtual worlds isn’t new, and at one time seemed like it was going to be the Next Big Thing. However that faded away, and apart from games like FUEL, the generation of worlds via computers, rather than humans is seen more as a starting point – something you then go in and polish and add real content to.
Now it seems that the whole idea is being left to hobbyists to explore, and perhaps come up with the “answers”. The best known example right now is probably MineCraft, where the area surrounding the player is built almost entirely procedurally. This has probably inspired at least two of the three I’ve been reading about. However, regardless of what inspired them, they all have different approaches.
Firstly, we have Shamus Young at Twenty Sided Tale, with his Project Frontier world. His goal is to create an interesting, varied world using procedural content. He’s not aiming for photo-realistic worlds, but rather a more cartoon style. He’s gone down the path of height maps (an image where each colour represents the height of the terrain), with various objects (such as grass, trees etc) added. What he’s done so far is quite beautiful, and I’m hoping some kind of playable game will come out of it. But then, I’m greedy.
Shamus also pointed his readers at Michael Goodfellow’s Sea of Memes, who is programming an MMO style game based on procedural content. He started with a “MineCraft like” world and branched out from there. So far he’s produced asteroids, floating landscapes, ‘tubeworlds’ and more. It’s been interesting to see the problems he’s come up against (which are different to those Project Frontier hit) and the possibilities for solving them. What’s most fascinating is that while he started with a MineCraft “cube-world”, he’s also exploring heightmaps and other ways of representing a world. I’m not really sure where he’s heading with all of this (and I’m not sure he is, either), but watching his project evolve is fascinating stuff.
The last blog I’m following is Procedural World by Miguel Cepero. This project is amazing for the level of depth and detail that he’s producing and aiming for. The landscapes (and buildings) are, even at this stage, incredible. His approach is probably best illustrated by the forests he’s producing. He doesn’t merely say “this is a forest area” and bang down trees. Instead he starts with a population of trees, and simulates them over time. This leads to a very realistic looking distribution of various types of trees, from those dominating at the canopy, to those hanging on to bare existence. He’s even been looking at simulating the political landscape in order to decide where cities and roads linking those cities will go. The whole path he’s taking excites me, and I really want to see where he ends up!
All three blogs have kept me interested and excited through the development of their various worlds. I really hope each one comes to a point where I can play in these worlds, in whatever game, or sandpit, they eventually evolve into.