Problem: Number of points generated from VIP doesn't change at small percent values
Using progressively smaller percentage values in the VIP command fails to reduce the number of points in the output point coverage.
The purpose of VIP is to create a point coverage containing the best possible set of mesh points to construct a tin. Two conditions will set the lower limit of the number of points that VIP will generate, regardless of the specified percentage value:
1) Before the points to use are selected, VIP includes all of the mesh points that define the perimeter of the data. For example, if the input lattice is a square of 100 by 100 points, there will be 99 (nrows * 2 + ncols * 2 - 1). This will also occur for any mesh points that are adjacent to NODATA areas defining the shape of the lattice.
2) VIP will also include in the output coverage all the perimeter mesh points corresponding to any 'holes' in the input lattice.
Once these points have been selected, the VIP algorithm proceeds to analyze the remaining mesh points according to the specified percentage.
The initial selection of point is discussed in the VIP documentation of ArcDoc.
Solution or Workaround
Smoothing the input data may reduce the number of initial selection points. You can accomplish that with either the TIN command FILTER or the GRID function FOCALMEAN.