[vtkusers] Re: vtkImagePlaneWidget Oblique Reslice Bug!?

[Istvan Matis]

Hi Dean,

I use vtkImagePlaneWidget with the version:
  Date:      $Date: 2003/02/04 16:40:47 $
  Version:   $Revision: 1.61 $

The exception was thrown in vtkImageReslice (used by vtkIPW) at line 1465,
sorry about that.
Here:
// pixel copy function, templated for different scalar types
template<class T>
void vtkSetPixels(T *&outPtr, const T *inPtr, int numscalars, int n)
{
  for (int i = 0; i < n; i++)
    {
    const T *tmpPtr = inPtr;
    int m = numscalars;
    do
      {
      *outPtr++ = *tmpPtr++;   //Here!!! //////////////////
      }
    while (--m);
    }
}

You were right , the two axes were not quite orthogonal (dot >> 0), I'll
have to work on that!

Thanks again!
Istvan

----- Original Message ----- 
From: "Dean Inglis" <dean.inglis at camris.ca>
To: "vtkusers archive" <vtkusers at vtk.org>
Sent: Tuesday, September 07, 2004 5:39 PM
Subject: [vtkusers] Re: vtkImagePlaneWidget Oblique Reslice Bug!?


> Istvan,
>
> I'm not sure which version of vtkImagePlaneWidget.cxx you
> are using, since
> "Undhandled Exception"s in vtkImagePlaneWidget.cxx line 1465
>
> line 1465 corresponds to a comment in the method:
> void vtkImagePlaneWidget::UpdateNormal()
>
> but, it if an exception is being thrown from UpdateNormal,
> it may be that you are not ensuring that points 1, 2 and the origin
> form part of a right-handed orthogonal basis.  vtkImagePlaneWidget
> does not check/ensure that those inputs mathematically make
> sense for reasons of efficiency.  Therefore, either the vtkMatrix4x4
> that receives the normalized vectors defining the directions along
> the new plane axes is causing an exception when it attempts to perform
> several math ops, or, the vtkImageReslice that receives this
> matrix is.  I would put a check in your code: e.g.,
>
> for(int i=0;i<3;i++)
>   {
>   axis1[i] = p1__[i] - o__[i];
>   axis2[i] = p2__[i] - o__[i];
>   }
> vtkMath::Normalize(axis1);
> vtkMath::Normalize(axis2);
> double dot = vtkMath::Dot(axis1,axis2);
> double nearZero = 0.0000001; // or some other very small number...
> if(dot > nearZero)
>   {
> //  ... DO SOMETHING APPROPRIATE HERE
>   }
>
> regards,
> Dean
>
>
>
>