I''ve been reading articles left and right about view culling, and the one thing that none of them seem to mention is whether they use an index buffer or not. As it seems, view culling algorithms cull each triangle(not each vertex), so I''m assuming that each triangle owns its own copy of shared verticies. But when it comes time to render, it seems like a waste because verticies that are shared by neighboring triangles are transformed more than once.(because each neighboring triangle owns its own copy of the same vertex) So if an index buffer is used, I can''t think of a good way to number the index buffer, since the indicies for the verticies in the vertex buffer changes according to the FOV of the camera. Is there a magical way that this is done? Or do people just take the hit with transforming verticies multiple times as a tradeoff for being able to do view culling?

I think you misunderstood my question. I''m wondering how you''d use index buffers when the verticies that you want to draw are changing all the time, since with view culling, you take a select portion of all verticies in the terrain.

Wil