[vtkusers] mixing vtkImageActor with solidly rendered vtkPolyData

[dean.inglis at camris.ca]

in a renderer I have a vtkImageActor (ivar of vtkImageViewer2)
showing a slice of a 3D image.  On top of that are 
two vtkActors rendering 2D contours (vtkPolyData lines) from a segmentation and
another actor rendering a 2D filled segmentation region (
vtkPolyData solid polygon).  When I scroll through the slices,
each slice's contour lines and polygonal region are
updated.  I can see the lines but the polygons appear
briefly and then disappear.  If I hide the vtkImageActor,
the polygons are clearly there.  If I set the polygon actor
to render as a wireframe, the region is correctly shown as a triangulated
polygon whether or not the image actor is visible or not.
If I toggle between wireframe and solid when the image actor
is visible, the solid polygon will not display: only the wireframe  
representation will.  SetCoincidentTopologyToPolygonOffset has no
effect.  How can I mix vtkImageActor with solidly rendered vtkPolyData?

Dean