I’m making a game with a friend using jme 3.1. It is for now a simple roguelite game played with an FPS camera. As we are both fond of ｖａｐｏｒｗａｖｅ, it is designed with the word retro in mind. Also, the project development is focused on the Agile & Scrum methodologies.
As in many roguelite games, the game heavily uses procedural generation for its levels.
I will mainly post media here and links to my blog. The purpose of this thread is to see the advancement of the game.
May I ask what algorithm you have used for the Voronoi diagram? Do you use Fortunes? Or do you go for more complexity than n log n? From time to time I scratch the topic Voronoi diagrams and I had some good lectures about it but I have never seen a working bug free implementation of the fortunes algorithm…
The library I use to create the Voronoi diagram uses Fortune’s algorithm. I then apply three times Lloyd’s relaxation algorithm, which is fortunately included in the library.
Well, in terms of code, it seems pretty bug-free. However, the controllability is not as I expected. I mean, I was able to retrieve each zone “center” point and each edge contrary to Joise which only generates pure N-dimensional noises including Worley’s noise. Nonetheless, I do not have a direct control on how many zones I have in a certain area; it’s actually an approximation from my part, especially because relaxing the graph 3 times extends the global area of all zones. Yet, I do have control of the total number of zones.
Recently, we’ve been working on a more classical approach to terrain generation : dungeons. Dungeon generation will be used for the jungle world. Right now, it generates rooms and corridors by crawling like a snake and thus having a maximum of 2 connections per area.