[vtkusers] vtkPolyDataNormals: normals pointing in the wrong direction

[Miguel Sotaquirá]

Hi David!

I'm not sure if the approach you suggested is adequate for my problem, since
vtkPointSetNormalEstimation has no concept of inside/outside object (which
in my case is necessary for computing curvatures)...

I don't know if vtkPointSetNormalOrientation in your code might help in this
case?
Thanks,
Miguel

2011/7/24 David Doria <daviddoria at gmail.com>

> 2011/7/24 Miguel Sotaquirá <msotaquira at gmail.com>:
> > Hi!
> > I'm using vtkPolyDataNormals to obtain point normals on my mesh. The idea
> is
> > to use these normals to compute curvatures using polynomial fitting.
> > The problem is that vtkPolyDataNormals is not giving me the correct
> normal
> > orientations for all vertices. Here's an example image:
> > http://tinypic.com/r/2dv7l36/7
> > as you can see, for now I'm using a synthetic, perfectly regular
> noiseless
> > mesh!
> > Here's my code for computing normals:
> > vtkSmartPointer<vtkPolyDataNormals> normals =
> > vtkSmartPointer<vtkPolyDataNormals>::New();
> > normals->SetInput( meshdata->polydata );
> > normals->SplittingOff();
> > normals->ComputePointNormalsOn();
> > normals->ComputeCellNormalsOff();
> > normals->Update();
> > Thanks,
> > Miguel
>
> I don't think there is any guarantee that normals computed by
> vtkPolyDataNormals are consistently oriented. A note to this effect
> should be added to the documentation.
>
> We wrote a EMST based normal orientation as a VTK filter here, it
> might be what you're looking for:
>
> http://www.midasjournal.org/browse/publication/708
>
> David
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.vtk.org/pipermail/vtkusers/attachments/20110724/fb03dcac/attachment.htm>