Can BJS not to render meshes which currently are not visible in a camera? So I think of separating one big mesh scene (buildings, floors, walls, etc) into some small pieces. Will it improve performance or not?
It will if you are GPU bound but it will not if you are CPU bound (as CPU will have more work to do with more meshes)
Babylon has deal with GPU when render for the most time. So separating one big mesh into smaller once makes common sense in terms of performance or it’s not an easy question as it seems?
It is not as I said: Are you mostly using the GPU or the CPU? But the universal answer would be to test and evaluate (I think you will probably end up with a big object)