I was unable to find any precise answer to this seemingly simple question. Even after watching the live stream about World composition, I’m somewhat confused.
They should have added a simple hierarchy diagram to the documentation to explain, how World, Levels, Layers, Landscapes, Maps, Tiles (landscape tiles, world tiles) relate to each other and what size and design approach is better for what purpose.
I thought there should be at least a few games that have the player navigating a small city, driving around and entering some buildings for a short moment (like a small cafe) or for longer periods of time (like a large building with many floors). But it seems I was wrong. And the authors of similar games (The Sinking City, for example) don’t reveal their “secrets”.
So for my purposes I’ll have to split the world into pieces and I’ll have to use world composition, which will take care to stream the levels in and out. But how large should these pieces be? Should I be ok with a single map of 5x5 km (most probably no) or 1x1 km or something smaller?
How to choose the right number of sections and components of the “floor” in the city and interiors?
And how to deal with entering / exiting buildings? Buildings have windows, so the outer world should be visible and a building might again be in the center of the world thus requiring loading of all 4 maps + interior map of the building. I know about culling volumes but not sure how to deal with them in buildings with windows.
It would be great to hear from developers who have worked on similar games or simulations and can suggest some default numbers and general approach that would work the best for my case.
BTW, in my case the player won’t be normally able to fly around, so it will be impossible to see entire world at once.
Of course, I’ll have to deal with LOD levels, disabled physics outside of max visible range etc. but I’ll deal with that later; at first I want to build solid and stable ground for my environment.