MtS - yes, I saw the code you posted. To be honest, you’re missing the point. Generating vertices from a heightfield map is trivial. You don’t seem to understand the kind of scales involved in high-quality terrain rendering.
Suppose your viewpoint is a human character wandering around on foot. For the terrain to look good close-up, you need a fine-grained polygon resolution, maybe a vertex every meter. But suppose you wander up to the top of a high mountain on a clear day, where you can see for 25 kilometers in every direction.
What are you going to do? Generate the entire 50km x 50km visible landscape at 1-meter resolution? Right, that’s 5 billion triangles. With vertices, normals, texcoords… call it 40 gigabytes of data per frame even with perfect tristripping. Hope you’ve got a fast system.
Obviously, I’m describing the extreme case here. You can probably get away with a lower resolution than 1 meter. You can probably get away with fogging at a closer distance than 25 km. You can certainly cull a lot of geometry that’s outside the view frustrum. I’m trying to get the message across that you CAN’T just bung out a 256x256 heightfield and expect to get good results. You’re off by a couple of orders of magnitude.
Take a look at Matt Fairclough’s Terragen (http://www.planetside.co.uk/terragen) to see what I mean by high-quality landscape rendering. Our mission, should we choose to accept it, is to get close to this kind of quality in realtime. Don’t hold your breath.