|
|
Well, I'm not sure I understood a large majority of the post, but I'll try and answer the best I can.
The main question: How do the pro's do it? Answer: Depends on which pro you talk to.
There really is no all encompassing "best way" of rendering large maps. (If there was, I think the programming would would be kinda boring.) If you look at some of the newer games out there, you'll find a lot of different methods.
Doom 3: A strictly portal based approach. This works really well for indoor arcitecture, but can be modified to work passably with outdoor scenes too.
Unreal Tournament (2k4): BSP based approach, but with "helper portals" throughout. Kind of a hybrid.
Quake 3: Straight BSP, with a brute force ray-trace visibility check run at compile time and stored in Potential Visibility Sets. Rather outdated approach.
Half Life 2: BSP based, if I remember correctly. I'm sure they have a lot of auxillary tricks to help with the outdoor scenes, though. Maybe an Octree?
It really depends on the type of game your looking at. If you are trying to create a game with a lot of indoor space, maybe a portal based implementation is best. If you want massive outdoor worlds, look into Quad/Oct-trees, along with some LOD algorithms. There's newer methods too, like occlusion culling using a pre-rended depth buffer (don't remember what it's called right now).
So what are you trying to create? That's the first question that needs to be answered. After that, the rest of the questions become a lot easier to answer. (Not to say that it's all going to fall into your lap, of course.)
|