They probably split the models of the terrain into different levels of detail, the further away the land, the less detail that has to be maintained. That way they can look farther and it still looks reasonable, but isnt taking up that much memory or processing time.

-Geoff

I'm writing a 3D program.
Right now I use a system of sectors. The camera is in the center. As it moves, the sectors behind the camera are dropped and new ones are added in front. This works, but it is slow and the screen pauses between sector loads. (It's only coming from memory!)
How do flight sims and the like have such big worlds? Is there a better way to do this?
The terrain is bumpy and can be looked from any height.

I understand how removing vertices reduces computational time a great deal. How do you handle large objects? For example, I have 5 points at altitudes:

1 1 1
1 5 1
1 1 1

That describes an entire sector. By taking away the point in the middle, I suddenly have flat land. By taking away any of the points around it, I suddenly have a large mountain. What is the answer?