Practically you'll just split your visible scene in slices and perform rendering for each slice. After that it is just a question to put those rendered slices together. Of course, you can and probably must use some scale tricks also with the further slices in order to keep the depth buffer happy.

Cheers!