Open RobertCochran opened 9 years ago
I was considering this. If you look at my algorithm, I do do this to an extent, simulating room-sections virtually (er... more virtual), then creating a new grid based on those somewhat arbitrary x's, y's, widths and heights. If we want to get really complicated, we could keep room in this virtual phase until we have arranged all the rooms, then create a grid with the appropriate rectangles. We'd need a lot of clever rules and collision detection methods for accomplishing this, though.
Another option I've been pondering involved establishing a virtual grid, where each section of the grid can have no more than one room within it. This could really simplify things. We'd have to adjust one of the existing room generating algorithms or create a new one to accommodate to a room size limit. But, I think this would all be far simpler in the long run.
Explain the whole 'virtual grid' idea. I'm not picking up on what you mean.
I think it'd be best if I we discussed it verbally, or if I programmed an example. Probably won't be able to do the latter, so we should try to set a date and time for a Mumble conference.
Maybe instead of trying to generate the world all in one go, maybe we should Instead have it generate separate 'puzzle pieces' that all fit together in the end. For example, one piece could have openings in the top and bottom, but then there are no requirements for inside the piece other than to be able to access all of the open edges. This piece could be combined with a piece that has an opening on the bottom and to the right, and so forth. This way we can keep a nice randomness, but then we have more freedom as to how the world is laid out.
Comments?