The plan was to draw a 256x256 map, and have a "portal" somewhere on the map that leads to another map. So loading all the maps when the game loads is out of the question.
My plan was to have a text file for each map with a whole bunch of vertices, followed by objects on the map with a position to draw it at.
map1.map, "Home Town"; 0,1,0,3,1,0,0,4,3,5,6,2,0,2,3,8,4,2,0,28,169,474,935,742,42,36; "House.obj", 56,0,56; "Car.obj", 56,0,80;
I have absolutely no idea how XNA loads objects, but this was my thought on how to do it.
Edit: Oh, and didn't need collision detection as I was just going to define each tile as walkable or not-walkable.
Does this sound like a reasonable, efficient way to draw a map?
Edit2: Actually, since the maps will all be 256x256, there's really no point in storing the x and z vertices in a text file since they will all be quads (tiles). The only thing that needs to be stored is the height of each group of 3 vertices. Which I suppose could be done with a height map image?