elevation map

Well, I have a lot of question about elevation maps, bc I want to use them to increase the level of detail of my ground.
First, I don’t know how fast it is. How many slices should I use to have good results?
Is it a good Idea, since not all card allow multi-texturing?
Should I use the RVB channel as a normal map, or as a color map? Then, how should I compute the normal map? Instead of using a normal map, should I use a simple light-map?
I only have read on tutorial (on Nvidia) about this technique, a great one, but without code sample. Where can I find other tutorial?
thx a lot.

It is typically faster to draw a coarse geometry and a second pass of pre-rendered shading than trying to crank up your geometry detail and/or do GL lighting.

Thus, even if you need to draw the thing twice, the second time with alpha blending for the map, it’ll be faster than the alternative for the same output quality. You may, however, want to detect whether the card supports the multi-texturing extensions and start using them if they’re available for best performance.

Yes, but the point is : if for exemple I use 5 slices (along Z axis) to render my elevation map, will it give enough details?? And will it slow a lot the rendering, if I use one pass multitexturing?? Because it seems to be much more intensive than a simple bump mapping…
And by the way, how do I create the normal map?? By rendering the surface with a high resolution??