I'm working on a 3D game with a large open world. It will naturally need to stream chunks. When you look at existing streaming solutions in the Asset Store, you can see that their approach is building the chunks as Scenes, and then additively load them.

That's fine, but in my mind it seems more logical to create and load the chunks as Addressable Prefabs, or perhaps, as individual AssetBundles. Yet, those assets seem to have been quite deliberate in this design decision. I've tried asking their developers, but their answers have been vaguely secretive as if there's some business logic to protect (e.g. "bundles have limitations").

I have come across this question, where the asker appears to be almost in the same situation: Unity prefab object limitations - however, they seem to be more concerned about the performance impact or the complexity of having thousands of objects inside a prefab (I don't believe this is an issue - to me a scene is virtually a prefab, except it has lightmapping features).

What are the technical differences (if any) between splitting the 3D open world into additively-loaded Scenes, as opposed to splitting the world into Addressable prefabs / AssetBundles?

As mentioned, lightmapping is the only notable difference that comes to my mind. Other than that, is there another technical reason developers appear to deliberately go for additive scene loading that I'm missing?

• Would the async versions of Addressable loading be functionally equivalent to scene preloading in this case? Aug 2 at 3:12