[vtkusers] Surface Normal Vectors

[Jim Peterson]

SHACKELFORD Steve (AREVA) wrote:
> Greetings,
>  
> I have a laser line scanned surface created using triangle strips.  
> When I interrogate the input vtkPolyData, I find that I have 107639 
> points.  When I create the vtkPolyDataNormals, I find that this 
> contains 108698 points -- there is obviously not a one-to-one 
> correlation between the input data and the output for the normal 
> filter.  I understand that the vtkPolyDataNormals filter takes some 
> liberties with the data to achieve a better visual presentation, but 
> what I am interested in is an accurate association of the normal 
> vectors with the original input data.
>  
> Any suggestions?
>  
> Regards,
>  
> Steve Shackelford
> Design Engineer
> Areva NP Inc., an AREVA and Siemens company
> 155 Mill Ridge Road
> Lynchburg,  VA 24502
>  
> PH: (434) 832-2424
> EMail: Steve.Shackelford at areva.com <mailto:Steve.Shackelford at areva.com>
> /**/ 
> //
I believe one reason for getting two (or more) normals for one point 
location is when the point is on an "edge" of the surface. I believe you 
can control the amount of smoothing using the feature angle parameter. 
My understanding is angles more acute that the feature angle will have 
cell associated normals to cause an edge to appear instead of smoothing 
the surface across the points. if the adjoining cells use the same 
normals for shared points the surface is smooth, if they have separate 
normals for the points an edge appears between the two cells.

Hope that helps,
Jim