[vtkusers] Design questions

[Roger Blum]

Hi,

I am about to start developing a vtk filter (my first!) to calculate 
viewsheds on a digital elevation model.

I have the following data:
- the DEM data (vtkImageData); float values
- Covering data (vtkImageData); integer values (covering codes)
- Covering heights (vtkFloatArray); a height for every covering code
- The observer position (3D vector)
- The observer height (Float)
- The target height (Float)

The viewshed calculates lines of sight from the observer position to every 
point of the given DEM, taking in consideration the observer height, the 
target height and the height of the covering. The output is an image 
(vtkImageData) with integer values (e.g. visible/not visible) for each 
calculated point.

I have a couple of questions for this problem:
- Has anybody done something similar before?
- Is it possible to do this kind of calculation with existing filters?
- Which class do I subclass (vtkSimpleImageToImageFilter, 
vtkSimpleImageFilterExample, vtkImageTwoInputFilter, vtkImageToImageFilter).
- Which methods of the chosen class need to be implemented / overloaded?
- Is it best to just link the vtk libraries to my project or is it best to 
integrate my class in a vtk library (if yes, which subdirectory)?

Any additional information which helps me develop my first vtk filter is 
very much appreciated.

Thanks in advance for your help,
Roger