[vtkusers] Re-using a vtkRenderer with a different image

[Charlotte Curtis]

I'm going to reply to my own question in case someone searches the
archive looking for the same answer.

There seems to be a problem with vtkImageData::PrepareForNewData().
In my code, if I try:

if (m_image)
       m_image->PrepareForNewData();
else
       m_image = vtkImageData::New();
m_image->DeepCopy(image);

The resulting m_image appears to be not completely overwritten, and
rendering is kind of funky.  However, simply calling Delete does seem
to work:

if (m_image)
       m_image->Delete();
m_image = vtkImageData::New();

I realize this probably isn't the "vtk way", but on the other hand it
works.  Could this be a bug in the PrepareForNewData method?

Charlotte

On Wed, Sep 2, 2009 at 4:32 PM, Charlotte Curtis<cfcurtis at ucalgary.ca> wrote:
> Hi all,
>
> I've got what's grown to be a rather complicated program based on
> ITK/VTK and wxWidgets (using the wxVTKRenderWindowInteractor).  If I
> just load an image (3D dicom dataset), all is well and it seems to
> display more or less correctly (in both axial and coronal views).
> However, if I then try to load a *different* image, it seems as though
> the renderer, or the image actor, or something is "remembering" values
> from the first image, and it doesn't display correctly - the coronal
> viewport is messed up due to anisotropic voxels, and the image appears
> "zoomed in" if it is larger than the first one.
>
> I've tried a number of things, like deleting and recreating both the
> imageactor and the renderer each time I load a new image, but that
> doesn't even seem to work.  Any ideas on what needs to be "reset" so
> that I can load a different image?
>
> The relevant portion of my code is as follows (where "image" is the
> new image, and m_image is the member variable holding the image in
> memory.  Similarly, m_actor, m_renderer, etc, are hopefully
> self-explanatory member variables).
>
> m_windowleveler->RemoveAllInputs();
> m_renderer->RemoveAllViewProps();
>
> if (m_image)
>        m_image->PrepareForNewData();
> else
>        m_image = vtkImageData::New();
>
> m_image->DeepCopy(image);
> m_image->SetUpdateExtentToWholeExtent();
> m_windowleveler->SetInput(m_image);
>
> if (image->GetDataDimension() == 3)
> {
>        switch (dim)
>        {
>        case AXIAL:
>                m_slices = image->GetDimensions()[2];
>                break;
>        case CORONAL:
>                m_slices = image->GetDimensions()[1];
>                break;
>        case SAGITTAL:
>                m_slices = image->GetDimensions()[0];
>                break;
>        default:
>                return false;
>        }
> }
> else
> {
>        int *extent = image->GetWholeExtent();
>        m_actor->SetDisplayExtent(extent[0], extent[1], extent[2], extent[3], 0, 0);
>        m_slices = 1;
> }
>
> m_range = image->GetScalarRange();
> m_windowleveler->GetOutput()->UpdateInformation();
> m_windowleveler->GetOutput()->SetUpdateExtentToWholeExtent();
> m_windowleveler->Update();
> m_actor->SetInput(m_windowleveler->GetOutput());
> m_renderer->AddActor(m_actor);
> vtkCamera *cam = m_renderer->GetActiveCamera();
> switch (dim)
> {
> case CORONAL:
>        cam->SetFocalPoint(0,0,0);
>        cam->SetPosition(0,-1,0); // 1 if medical ?
>        cam->SetViewUp(0,0,1);
>        break;
>
> case SAGITTAL:
>        cam->SetFocalPoint(0,0,0);
>        cam->SetPosition(1,0,0); // -1 if medical ?
>        cam->SetViewUp(0,0,1);
>        break;
>
> // Axial and single slice
> default:
>        cam->SetFocalPoint(0,0,0);
>        cam->SetPosition(0,0,1); // -1 if medical ?
>        cam->SetViewUp(0,1,0);
>        break;
> }
> double *spacing = image->GetSpacing();
> int *extents = image->GetExtent();
>
> cam->SetParallelScale((double)(extents[3]-extents[2])*spacing[1]/2.0);
> m_renderer->ResetCameraClippingRange();
> double scale = m_renderer->GetActiveCamera()->GetParallelScale();
> m_renderer->ResetCamera();
> m_renderer->GetActiveCamera()->SetParallelScale(scale);
> m_iren->Render();
>