[Paraview] choppy derivatives

[Moreland, Kenneth]

If you are looking to try different derivative filters, you could try the one of the Gradient filters.  They are similar to the ComputeDerivatives filter but should allow you to combine steps 2 and 3 together.  There is also recently added a vtkDataSetGradient filter to VTK.  It's not exposed in the ParaView filters list, but you could do so by plugging in some XML specification.  It provides yet another way to take the gradient and can go directly from cell data to point data.

-Ken


On 3/27/10 11:45 PM, "Peter Brady" <petertbrady at gmail.com> wrote:

Hello list,

I have a 3D dataset with a Temperature field.  My goal is to insert a sphere source and determine the tangential temperature gradient on the surface of the sphere at various zenith angles.  The data is output from the code as cell data.  Here is the way I'm doing this (through pvpython)

1.  Apply CellDataToPointData
2.  Apply ComputeDerivatives (to get the 'Scalar Gradient')
3.  Apply another CellDataToPointData (to turn the derivatives into point data)
4.  Resample the data from step 3 with my sphere source.
5.  Use a programmable filter to combine the xyz derivatives appropriately.

The data is reasonable but it's fairly choppy.  When I look at dT/dx, dT/dy, and dT/dz (before I project them onto the sphere surface), and plot them at a particular zenith angle around the azimuth of the drop, dTdx and dTdy are very smooth but dT/dz is fairly choppy (especially above the equator of the drop).  If I expand my sphere to a place where there's not a lot going on dT/dz seems to calm down.  Is there a way to smooth dT/dz in the location of interest or should I try to write my own derivatives filter to make them smoother?  Any thoughts?

Thanks,
Peter.



   ****      Kenneth Moreland
    ***      Sandia National Laboratories
***********
*** *** ***  email: kmorel at sandia.gov
**  ***  **  phone: (505) 844-8919
    ***      web:   http://www.cs.unm.edu/~kmorel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.paraview.org/pipermail/paraview/attachments/20100329/61162a73/attachment.htm>