Greedy Insertion Parameters
The --tc-strategy greedy method implements a slightly modified version of the algorithm described in:
M. Garland, P. S. Heckbert, Fast polygonal approximation of terrains and height Fields, Technical report CMU-CS-95-181, Carnegie Mellon University 575 (September 1995).
The method starts with a very rough triangulation of the data in the XY plane, which is typically the two triangles resulting of triangulating the four corners of the tile in the XY plane. Then, for each of the remaining points from the initial sample we compute the error from this approximation to the points. At each iteration, we select the point inducing the largest error and insert it in the triangulation (Delaunay insertion). Given the newly generated triangles, the errors for each of the remaining points are recomputed, and the insertion process is repeated until all the errors are below a given threshold.
The parameters ruling the method are the following:
-
--tc-greedy-error-tol(*): Error tolerance for a tile to fulfill in the greedy insertion approach. -
--tc-greedy-init-grid-size: An initial grid of this size will be used as base mesh to start the insertion process. Defaults to the 4 corners of the tile if < 0. This parameter is useful for having some complexity on tiles on shallow zoom levels (0..4) if we want to visualize this data in spherical coordinates (as it is usually the case in Virtual Globes) -
--tc-greedy-error-type: The error computation type. This parameter is experimental, and we do not recommend changing it from its default value ("height").