[vtk-developers] imaging pipeline bug

[Amy Henderson]

Hi Dean,

I'm seeing the incorrect results you reported: N: 0, (1E+299 -> -1E+299) 0 
+- 0 HU

My VTK (latest CVS version) is built with shared libs using MSVC++ .Net 
7.1, CMake 2.0.5, Debug, Win XP Pro.

This should probably be added to the VTK bug-tracker.

- Amy

At 02:48 PM 1/3/2005, Dean Inglis wrote:
>Hi Amy,
>
>thanks for trying the script  This is odd.
>I get a crash on shared lib build under Borland C++ Builder
>vers. 5, CMake 2.0.5, MinSizeRel, Win 2000 Pro, latest cvs of VTK,
>but no crash on a static build on a different PC (same software).
>However, on the static build, the script runs showing the correct
>rectangular ROI but the image stats are not correct:
>the vtkImageAccumulate calculates all zeros.  I successfully ran the
>script under a static build with revision 1.51 wherein the
>ROI stats are all non-zero (i.e. correctly calculated).  So, if you are
>you seeing the text:
>N: 0, (1E+299 -> -1E+299) 0 +- 0 HU
>
>as I am, then something is wrong. The correct numbers should read:
>N: 1504, (100 -> 2317) 1051 +- 419.018 HU
>
>Also, this may or may not be a separate bug with vtkImageAccumulate,
>but if the ivar VoxelCount is 1, a divide by zero will occur:
>see line 244 in method vtkImageAccumulateExecute.
>
>Dean
>
>
>
>_______________________________________________
>vtk-developers mailing list
>vtk-developers at vtk.org
>http://www.vtk.org/mailman/listinfo/vtk-developers