Is it a good idea to use LODs without actually using lower detail meshes, thus only having an LOD0 and a culling state ?
and would it still be beneficial if you have a lot (100’s) of very small objects all having this LOD component just too cull it when it’s far away ?
so basically I’m just wondering if I should use LOD on every object in my scenes (1000’s of objects), or just the biggest meshes ?
btw, the scenes in the project I’m working on are really big, and very open, with lots of small objects, these scenes cost a lot of memory and are quite expensive.
I haven’t directly futzed around with Unity’s LOD systems too much, just yet, but what you’re talking about sounds like a familiar pattern across many engines: avoid drawing “detail” meshes at long distance, because they’re a waste of precious render time if the player can’t see them. For example, Unreal gets very explicit about this with size-based distance culling.
In theory there’s a small memory/CPU hit for this sort of thing, but it’s usually negligible when compared with the potential render cost you’re saving. It might be worth profiling the project a few times as you add LOD to more and more objects, just to be sure of exactly what you’re saving (or not). I imagine it’ll always be saving, but it never hurts to know for certain.
Depending on the game, frustrum and occlusion culling can usually be very helpful as well.
For performance reasons, you might want to cull small objects earlier. For example, small rocks and debris could be made invisible at much smaller distance than large buildings. To do that, put small objects into a separate layer and setup per-layer cull distances using Camera.layerCullDistances script function.
Normally Camera skips rendering of objects that are further away than farClipPlane. You can set up some Layers to use smaller culling distances using layerCullDistances. This is very useful to cull small objects early on, if you put them into appropriate layers.