Jordan Fisher, founder and lead programmer of Pwnee Studios, developer of AI-generated and -tested platformer
Cloudberry Kingdom,
explains four things that are gained by procedurally-generated levels.
Cloudberry Kingdom was inspired in part by the
Mario series, well known for its handcrafted level designs. How can a procedural approach stack up?
Here are four advantages Fisher outlines:
Challenge. No matter how good a player gets, the algorithm can make a level that will challenge her. Likewise, if a player is more casual, the algorithm can cater to her skill level.
New game types. We can make game types that depend on random levels, rather than just utilizing them. For instance, imagine an infinite string of levels strung together in a
Tetris-like progression of difficulty. Without randomness, this game would devolve into a memorization challenge. With randomness, the focus shifts toward the player's skill, rather than muscle memory.
Customizability. Players can tinker with the algorithm's parameters, creating levels built to their specifications. There is a vast universe to explore.
Uniformity. This is opposite of customizability, but it's an important tool. We can fix the algorithm's parameters but still get an infinite string of levels, all very similar in feel, but different nonetheless. For instance, we can fix the "flow" of a level, so that each level can be beat without pausing or backtracking. Maintaining this consistency can help a player get in the zone, although it has to be balanced against trying to keep the player interested.
The full feature, in which Fisher explains the ins and outs of the design of his level generation and testing mechanism, is
live now on Gamasutra.
