Dan,
Thanks for your input, which helped us to realize what we did wrong:
First, we should do this in ArcScene, rather than ArcMap, though 3D analysis is still available in the latter;
Second, for both buildings and points data, we should run a function 'Feature to 3D by Attribute' in 3D Attribute toolset. This will change both data from 2D into actual 3D data;
Third, then we run 'skyline' using these two converted 3D data. This time, we could see the appropriate skylines generated.
Some more comments/thoughts/queries:
(1) the range of skylines is determined by the values in 'Virtual Surface Radius' (make sure to use the same unit as that in your data);
(2) if you want to use the skylines to calculate skyline graph, then the range of skylines will not affect the calculation of angles;
(3) currently, we simply assume that the base level of the point data is the same as that of the buildings, e.g., if the height of a point is 0, then it meanse the skyline from this point starts from 'ground' on which the buildings are based. We are not sure if this is a correct assumption. Also, this ignores the impact of terrain. We have DEM of the same area, but we are not sure if we use DEM to define 'Input Surface', how should we adjust the base level of both points and buildings. Should we simply add the elevations at their locations to points and buildings? I thank anyone who may provide some insight.
Peng