[vtkusers] A bad visualization idea?

[andrea_gavana at tin.it]

Hello NG,

    what I am basically trying to do is to visualize 3D 
grids like this one:

http://xoomer.virgilio.it/infinity77/grid.JPG

This picture is a result of this sequence:

1) 3D finite difference 
approximation scheme (grid) for numerical simulation (with 100 cells in 
x-direction, 43 in y-direction and 23 in z-direction);
2) Filtering of 
inactive cells via vtkThreshold

Well, vtkThreshold outputs a 
vtkUnstructuredGrid, to my knowledge. In that picture, the grid has 
about 80,000 cells (hexahedrons). It is not a big grid in the numerical 
simulation fields.
My concern is that handling bigger grids becomes 
almost impossible because the rendering speed decreases a lot. 
Rendering and interacting with a 200,000 cells unstructured grid is a 
pain on a 1 GB Windows machine.
I always thought that 3D finite 
difference schemes where displayed using a vtkStructuredGrid, but by 
looking at the image I can't imagine how to setup a vtkStructuredGrid 
to represent that.
This is not a complain about VTK, I am happy to use 
it... however, all our commercial visualization tools can handle also a 
million cell case without a blink, and I am wondering if the strategy I 
am applying is wrong or probably VTK is not the appropriate tool for 
the job...
Should I take another approach and not using 
unstructuredgrids? Are there better strategies to view 3D finite 
difference grids? Does anyone have some suggestions?
BTW, I am using 
VTK 5.0 with Python.

Thank you very much for every idea.

Andrea.