So I'm working on a simple prototype that I want to use to learn about 2D random map scripting. My ultimate goal is to include random map scripting in my current project, but before I get that far, I want to make sure I've got the basics down. Unfortunately, I just spend four hours struggling to get a script that evenly distributes X players on a square map, and that's probably going to be the least of my troubles.
I've looked around on the Internet and found lots of articles concerning 2D cross-section fractal maps (à la Scorched Earth), 3D mesh generation, and Perlin noise, but none of these seem helpful. Perlin noise looks popular for some applications, but it's very blotchy-looking and I'm not sure I can effectively use it to control the various parameters of the map. Strangely, I've found very little documentation on the algorithms used in games like Civ4 or AoE (Age of Empires II being the particular game I want to emulate with this prototype). I saw a single mention of flooding algorithms as being the basis of the RMS in Age of Empires, but it was only a fleeting mention and, having read the Wikipedia article on flooding algorithms, I'm not at all sure how they can help me in writing a random map script.
Can someone give me a basic breakdown of what a random map script would look like, one that can accept variable parameters for map size, number of players, number and width of rivers, lakes, number and size and spread of patches of trees and so on? Terrain height is not something I want to get into for the purposes of this prototype.
Also, as a side note, what kind of mathematical/computational algorithms can you use to evenly distribute points throughout a given area? I was trying to get my program to distribute X points (representing player start locations; X would be between 2 and 10) along the edges of the map, being located between 5 and 10 tiles from the edge. It was extremely difficult and the end result still isn't as tidy as I'd like it to be around the corners. I'm sure this has been done a lot, though, and much better than I have done.
FreeCiv is open source and it has random map generation with a lot of parameters that let you control various features. You could try taking a look at how they do things. Unfortunately, it's probably hard to understand, since it's much easier to write code than to read it.
I trust exceptions about as far as I can throw them.
It depends on what exactly you're trying to create, the scale of the map and the details on it, but in general it's good to view the procedural map creation process as a series of brushes and filters.
For example, creation of continents can be done by using a fractal image of a resolution higher than the grid resolution, and then by sampling various parts of the image into corresponding land/sea tile types. Then, you can apply various image filters (on the black/white aka land/sea image) such as blur (to smooth out the coastline). You can change parameters of the fractal generation to clip at different levels, or create one image for each continent and then merge them together with a blend map. For archipelago you can use perlin noise, for example. In general you first create the land/sea map, and then you fill out the land.
Giving different terrain types to land can also be done in various ways. For example, you may decide you want hill chains on random areas around the map. You can then create a curved line (spline) from one point to another, or more curved lines. Then you can attach a circle to that line, and start with the circle being a small radius, then it increases, and then decreases when it gets towards the end of the line. Then all the tiles within the circle become hills. However that on its own would seem too uniform, so you can apply a substraction filter by using perlin noise on the edges of the hills area to fill it out with different types of terrain. You can control the size and shape of each of those types by making each brush and stroke take parameters. This is just one method to give you an idea in which direction you could be heading. There are of course many other approaches.