I have two implementations of my terrain project, one for SL3+ and one for SL2 for both HiDef and Reach modes of XNA. I want to support Reach really just for the joy of it (its a hobby, not commercial).

The HiDef mode uses a texture heightmap and a quadtree of tiles, all sharing the same mesh and skirts which I scale as needed.

The Reach mode doesn't support texture lookups in the vertex shader, so I've got seperate meshes for each of the tiles in the quadtree, which I load as needed.

Can anyone suggest a way of optimizing the Reach mode to use less Vertex Buffers and therefore less resources, akin to the heightmap based one, but compatible with SL2 ?

A further annoyance which hadn't occured to me when going down the "fixed mesh" route for Reach support was the limited nature of the scaling - essentially by adding in a standard wave function (encoded in a texture) I can make the HiDef mode scale to a very high number of triangles at very close range, but with the Reach mode those need to be encoded into the Vertex Buffer meaning high level of detail of the near-to-camera mesh is too resource heavy, and I end up with an obviously triangle based near-to-camera level of detail.

Thanks for any suggestions;

Phillip