Getting Clear on Occlusion Culling

Hello,

I have been using the occlusion culling, and its been working, I just wanted to get some clarification and further input.

First of all, I believe the following items are true

  1. Occlusion culling will not split a mesh
  2. Really, to use this and not have it look weird, you need barriers, or closed doors, or some kind of barriers in your level you can cull otherwise items will come in and out of view as if from nowhere.

Based on the above, and knowing that (roughly) iPhone should stay under 11K verts with as few drawcalls as possible …

Do the satements below seem accurate …

Before culling: Really limited to changing camera setting to not render large areas, overall game level verts needed to be very low.

With culling: I can use a game level with far more verts as long as I have barriers and split meshes or objects and cull in the viewable areas with ‘View area’ squares through culling.

Is that the correct way to do it?
I have now seen two videos, one just culls the entire level and uses very small squares, others say cull the actual areas (rooms, etc…) with much larger squares.

(I am referencing the youtube video that uses 100s of small squares, and the unity docs that use larger squares in specific areas.)

Thank you for your feedback, and opinions on this.
It’s crucial that i understand this and its best practices, i’m just making sure.

Thanks again,

Trevor

  1. Correct, occlusion works on per object base
  2. This highly depends on the model. normally it shouldn’t pop in half in screen already. if that happens your cell size was not appropriate for the object size you have (too large cells).
    I would commonly recommend to keep it in the range of 1 - 2 times the smallest dynamic object you want to be occluded.

Other than that you are right.