Q: Is there any dependence between texture and geometry databases in the "Hardware Independent Clipmapping" technique?
A: Most terrain visualization techniques have a strong coupling between geometry and texture databases. It is necessary to subdivide the texture in order to adapt it to the geometry or vice versa.
Our technique stores the image in tiles that are not used directly as textures. These tiles are combined in a texture stack that caches the region of interest, following the clipmap idea.
We don't need to subdivide the geometry to make the patches match the texture tile boundaries, as occurs with most terrain visualization techniques. Whatever the geometry algorithm, there will always be a texture to map each patch.
The only drawback we have is that the bigger the geometry patches, the coarser the texture level applied. When a geometry patch exceeds the boundaries of a texture level, the level cannot be applied, switching to the next coarser level. This problem can be alleviated increasing the clip size, so the texture levels cover a more extense terrain area.
So the limitation of our technique is more about patch size than geometry structure, subdivision or tesselation.
The implementation of our texturing technique has been successfully used with different geometry algorithms, based on grids as well as TINs, with some slight level dropping when using large geometry patches, which can be usually avoided.
Anyway, this limitation is solved in our new texturing technique that also adds some new functionalities.
Last modified: February 18 2009.