Hi all,
Right now I’m trying to optimize my level. I was wondering how everyone usually modeled the ground floor for their levels? Do you use one continuous model? Or split that model into chunks?
Hopefully splitting the ground into chunks will help bring my performance up because my numbers are pretty bad, haha.
It depends entirely on what sort of game it is.
If you are looking to produce many rooms within a house it would be smart to divide it into each room, as occlusion culling would optimize it very well.
However in such instances as a large terrain you might want to make it a bit larger, but still split it up. Fog can seal off any far clip plane optimizations and add great effect to the game.
In the end it depends on how much of the level people are seeing. You should be trying to give them as much detail as possible while showing as few verts/drawcalls as you can.
It’s a sidescroller so the view is pretty limited. I figured keeping the ground floor one continuous model would reduce draw calls needing to draw each chunk of the floor.
If the floor is big (many times the size of the screen), you would expect to get better performance by breaking it up. Reducing draw calls is all well and good, but you can avoid drawing things altogether if they can be culled. I would guess that an area slightly bigger than the screen would make a good basic tile for the floor.