[vtkusers] vtkClipPolyData: problem computing output bounds

[Sebastien BARRE]

At 04/12/2000 11:22, Audrius Stundzia wrote:

>I've run into a puzzling problem using vtkClipPolyData.
>My input is a vtkPolyData surface which I left-right bisect using vtkPlane 
>and vtkClipPolyData. The number of triangles
>for each side is about half of the whole surface, as expected.
>
>Here's the problem: computing and getting the bounds for the left and 
>right bisected surfaces returns the bounds of the input whole surface for 
>each of the left and right bisected surfaces, which is clearly
>wrong.

Do not forget that for efficiency reasons many filters share the same 
ressources : for example, the same reference to the point coordinates.

In the latest official release (3.1), the bounds of a polydataset were 
computed from the set of associated points, thus implying the participation 
of the points that might not be *really* used by the topology defined in 
the dataset. I've fixed that behaviour some time ago so that GetBounds() is 
computed from the points coordinates involved by the current topological 
relationships. It's in the CVS, thus will be in the new soon-to-be-released 
VTK official version.