[vtk-developers] Re: [vtkusers] contour lines not closed with vtkUnstructuredGrid

[David C. Thompson]

> Hi, I have an issue with contours of some multiresolution data. The data is 
> comprised of adjacent "boxes" that don't overlap represented using 
> vtkImageData. ... The 
> problem I'm having is that contours of the resulting dataset aren't closed, 
> and I can see that where they separate is where the grid spacing changes. I 
> think this is a result of less acurate interpolation in the coarser spaced 
> regions. even so shouldn't the contours be closed? Is this an issue with the 
> contouring algorithm in vtk or do I have to admit defeat and forget about 
> visualizing multi resolution datasets?
This is not an error. It is a property of the interpolants you are
using. If you have a large cell dx=3, it might have endpoint values of
f=0 at x=0 and f=3 at x=3. But a neighboring region with dx=1 might have
values of f=0 at x=0, f=2 at x=1, f=3 at x=2 and f=3 at x=3. If you ask
for a contour showing where all cells take on the value x=1.5, this
occurs in the large cell region at x=1.5. However, in the small cell
region it occurs at x=0.75 (between nodes at x=0 and x=1). But why would
discontinuities in your isocontours cause you to admit defeat and stop
visualizing your solutions? The solutions weren't any better before you
drew pictures of them. :-)

	David