[vtkusers] vtkLandmarkTransform doesn't work as described

[jimmigoo]

Hi VTKUsers! 

I need to obtain an affine (or pespective in a better way) transformation,  
starting from a vtkPoints (or vtkPolyData) set of points ( i.e. 3 points
chosen by user ), 
to a vtkPoints (or vtkPolyData) set of other points, again chosen by user). 

I found that vtkLandmarkTransform seems not work if I use SetModeToAffine()
method
(I obtain a NaN matrix with GetMatrix() method), that's why David Gobbi said
to me that
it requires that all points are non-coplanar. So, I tried to insert the
source point set and
the same ones with different z-coordinate (I work on 2D images, i.e. aerial
photos). 
In this situation, a homogeneous matrix was built, but with the fourth row
(representing the
perspective ratio), set to 0 by the vtkLandmarkTransform itself. 
 
Therefore, if I leave default settings (Similarity), the matrix obtained
doesn't transform other
vtkPolyData set into the right ones. 

I tried to use vtkIterativeClosestPointTransform instead, obtaining not good
results.  

Does anyone have an example (even using the vtkINRIA3D classes instead of
the original VTK ones),
to understand how to correctly use those classes. 

I'm look forward for a fast reply.
Cheers, 

jimmigoo
-- 
View this message in context: http://old.nabble.com/vtkLandmarkTransform-doesn%27t-work-as-described-tp26560085p26560085.html
Sent from the VTK - Users mailing list archive at Nabble.com.