Ok, I see you’re poking around. Cool. I have no plans yet for mapgen5 but I have some experiments where I learn new techniques that I might be able to use.
I’d like to have more interesting terrain: chasms, cliffs, bays, lakes, estuaries, river deltas, plateaus, etc. I’d like the angle of repose of snow (which varies by temperature!) to affect whether mountain tops are rocky or snowy. It would also be interesting to add features on top of the map: towns, roads, nations, trade routes, natural resources, etc. I think drawing bushes and trees would make things look much more interesting.
In mapgen4 I pushed myself to use a very small procedural generation “budget”. The entire simulation (elevation, mountains, valleys, evaporation, wind, rain, river flow) and the entire 3d rendering had to fit within 1/30th of a second. I recalculated everything from scratch every time you painted even a pixel. In optimizing that code I made it much less flexible, and much harder to work with. For mapgen5 I think I’d like to have a much larger “budget”, maybe 1 second or more, so that I can run much more interesting procedural generation algorithms, make the code much more flexible so that it can be adapted for use in many projects, and make the maps much larger (mapgen4 uses 25,000 polygons but I’ve tested as many as 600,000).