My game has a simple background which is currently a single large mesh (around 2000 polygons). Performance wise is it better to split this up into several different game objects? I’d heard that Unity culls geometry by culling game objects against the view frustum and not at a polygon level using an quadtree etc so i’m thinking it would be able to do a better job at culling if the background were split up. Although this would create more draw calls.
Yes its much better.
Culling is done on per object base, not on polygon level.
optimally you make the submeshes that large that they cover something around 1.33 screen sizes, that way the common situation will only have 1 or 2 of them on screen. 4 is rare and more is impossible at all
dreamora, thanks for your reply. Useful to know. Traditionally i would use quadtree as it’s easier for artist to build the level as a whole.
Thats naturally a way as well.
Problem just is that the iphone isn’t exactly a calculation monster and quad trees aren’t exactly lightweight
So you would have to benchmark the exact use of full dynamic quad / oct tree against a static multimesh approach