Before a point of interest is selected, the user is asked to decided on the block size of the subgrid. The subgrid is used to exploit the LOD node's view volume culling and multiple resolution representation. As a result, the user's browser can provide the best presentation of the data. In the future, these selection should be made more transparent.
The polygons within each element of the subgrid is optimized. The edges of the grid are extracted at the highest resolution to ensure the continuity between grid elements. As a result, a large grid element allows better polygon optimization. However, since large grid element size implies less LOD nodes, the browser has limited options on its presentation.
Two set of VRML terrain files, test14.tar.gz and test15.tar.gz , are available form anonymous ftp. test14 shows Albany, NY. The terrain is relatively flat, so there are only a small number of polygons. test15 shows a hilly area about 82Km north-west of Albany. You can see the difference between different levels of detail with this model. All these terrain are available form the following clickable map interface. The file with a .out extension contains some information about each level of detail.