My latest game will take place on a small planetoid. I am looking for good data structure for representing cells on the surface of a sphere. Triangles, squares, pentagons, hexagons? Which one minimises stretch the most and creates the best tiling?
Spherical mapping is the easiest but the stretch at the poles is unacceptable. Cubemapping is also fairly easy but there would still be considerable stretch near the cube corners. Subdividing an icosahedron seems the best in terms of stretch but there is the problem of indexing many triangular arrays and finding neighbouring cells at the boundaries would be difficult.
I guess I could use a single linear array of points representing N-gons, each with an array of N neighbour indices, but that seems like a huge waste of space.
The game has RTS elements so I will be storing things like influence maps and performing A* pathfinding and convolution, so the representation has to be efficient.
No comments:
Post a Comment