[vtkusers] Gradient estimation in volume rendering

[Benjamin King]

Hello,

I have some difficulties in understanding how the different volume 
mappers deal with the gradient opacity transfer function.

In a nutshell, my questions are
1) Does VolumePro use the whole range of the gradient magnitudes or are 
they clamped in some way?
2) How can I provide the VolumePro card with my own gradient magnitude 
volume?


As I understand, the vtkVolumeRayCastMapper uses an internal 
vtkFiniteDifferenceGradientEstimator if the user doesn't provide her 
own. If no shading is performed, only the gradient magnitude is 
considered and it has 8 bit precision. The scale and bias are left on 
their defaults, 1.0 and 0.0 respectively, so the gradient magnitudes 
above 255 are clamped to 255. But I have datasets with a larger range 
gradient magnitudes. How does the vtkVolumeProVP1000Mapper handle this?

And my second question arose, because I already have to compute the 
gradient magnitudes of the volume for some other purpose. It would be 
consistent (and possibly faster) if the VolumePro would use them instead 
of recomputing them. The 'VolumePro 1000 Principles of Operation' manual 
says, that you could provide the card with your own gradient 
information, but I don't know how to do this with VTK. If it would be 
possible, I also could try some other scheme of gradient computation 
than central differences as well.

Has anyone an idea or a good source of information?

cu,
   Benjamin King