Implemented the base map of the level using the principles of the wave function collapse algorithm alongside using weighted directions to make the generated level less maze-like.
Separated out portions of the map by the three development phases, color coded for clarity.
Removed tiles that would be not traversable to a player as they are going through the map.
Utilized a drunken walk to simulate rivers between the islands and distinctly divide the phases for clarity.
Updated the algorithm to spawn bridges and boss rooms at choke points between the phases.
Counted tiles and other data to ensure phases were not too short and to also ensure they were engaging enough (see previous development phase 2).
Added a chance to clip corners of the map to have the first and third islands feel more natural overall.
Developed a simple and easy pipeline for allowing other designers to get the modular level prefabs into the corresponding spots for generation.
Completed a second pass of the map to add decorative edge tiles, including replacing non-traversable areas with these tiles as well.
Finished generation with the tutorial and the boss area added to the map.
Once the algorithm generates, but before spawning any objects, I had used a breadth first search queue structure to ensure the level is possible. In addition to this, it also finds an optimal path from the beginning to the end, which updates signifiers within that path to help guide the players.
Gif showing the different generations of the 3 development phases, helping fulfill the game pillar of "Everchanging Islands of Discovery!"
For Cats N Critters, I collected telemetry data in order to fine tune level sizes, targeting an optimal-path gameplay experience of around 15 minutes. This system allowed me to get information from 134,000 iterations while I was away from the project overnight. This data-driven approach helped balance gameplay pacing and provided insights into how well the wave function collapse algorithm performed, allowing me to identify and resolve bottlenecks to enhance overall game flow and algorithm efficiency.
©2024 All Rights Reserved By Corbyn LaMar