[vtkusers] vtkImageReslice and slab mode

[David Gobbi]

Hi Sankesh,

When I wrote this filter, I decided not to reduce the extent.  Similarly,
if someone uses vtkImageReslice to apply cubic interpolation (which has a
kernel size of 4), I don't remove two slices from each of the 6 faces of
the volume.

Instead, for any kernel-based operation, if any of the kernel samples are
outside of the input volume, then the default behavior is to clamp the
coordinates of those samples to the bounds of the input (the same as what
OpenGL does when you apply CLAMP_TO_EDGE with a texture lookup).  The
vtkImageInterpolator also provides other methods for handling the boundary.

So, in the case you describe, the slab at the boundary will contain repeats
of the outer slice.  To avoid this, you'd have to explicitly set the
OutputExtent yourself, or else override RequestInformation().  It used to
be possible to wire such changes into the pipeline easily via the
ExecuteInformationEvent.

Really, a way to customize RequestInformation() is what is needed here...
almost everyone seems to want something different, depending on their
particular application, and the API for vtkImageReslice is already waaay to
complicated.

 - David


On Mon, Jan 30, 2017 at 8:00 AM, Sankhesh Jhaveri <
sankhesh.jhaveri at kitware.com> wrote:

> Hi David,
>
> I am looking at the vtkImageReslice class to reformat a vtkImageData,
> specifically at the slab operations to generate thick slices.
>
> When I set SlabNumberOfSlices to be greater than 1 and SlabSliceSpacingFraction
> = 1, I’d expect the output dimensions along the reslice direction to be
> less than the input dimensions. But, they come out to be the same. What am
> I missing here?
>
> Thank you!
> Sankhesh
> ​
> --
> Sankhesh Jhaveri *Sr. Research & Development Engineer* | Kitware
> <http://www.kitware.com/> | (518) 881-4417
> ​
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/vtkusers/attachments/20170130/2ffa8eb7/attachment.html>