Loss of precision in many VTK filters
This issue was created automatically from an original Mantis Issue. Further discussion may take place here.
Many of VTK filters that work with vtkPolyData create vtkPoints for the output as:
newPts = vtkPoints::New();
This produces float vtkPoints. If the input vtkPoints are of type double, the filter converts double point coordinates to float, with the devastating loss of precision for high dynamic range data. I think you need to go over all chidren of vtkPolyDataAlgorithm and replace the line above with
newPts = vtkPoints::New(inPts->GetDataType());
Without this fix, many of VTK filters are unusable for visualizing Adaptive Mesh Refinement data, which may easily have dynamic range of more than the precision of float, 10^7.