Impostors, and allowing visualization from above. Also, a fallback for “merging” several maps together visually without using impostors.
It depends on what you do. If you try 12k tiles. (1km per tile, 110km^2 area) you just end up overloading the system with drawcalls.
you can cover the same area with larger maps. 8km^2 each (169 tiles approx) and gain tremendously on performance for impostors.
Overall, the default world composition system doesn’t really pair well with 1:1 areas and realistic eye view distance.
A viable working alternative is to have a very large mesh that covers the 11km at a total of vertex that is below the sum of the vertices of all the impostors.
1 draw call. Almost always rendered in the scene.
A performance hit to he sure, since its probably Double the cost of the tiles in terms of vertex count - possibly more. Depends on field of view setup/fustrum culling.
You can theoretically section the gigantic mesh off like you would any other sections to help with occlusion.
(My biggest problem is that even with 110km you are still missing areas that would be visible IRL to the eye)
Either way, there are situations where a large whole can be beneficial.
@OP
Sounds more likely that your pc/rig is limiting your creation process.
On a top down game you have a tremendous amount of room for optimization given you know how much/little of the world composition slices will be shown depending on camera setup.
however to actually make the game you likely still need 64gb of ram and 12+ of vram on the same gfx - sli doesn’t work for development.
Ofc, if you hit a limitation on that kind of specs, you need to definitely start to unload some levels…