[vtkusers] about vtkStreamer

[Petru Pau]

Jeff,

Thank you very much for your offer to help.

I've just had a short look at vtkStreamTracer. The approach you suggest
is somehow different from what I did for vtkStreamer: I projected the
velocity vector on the surface of the current cell, and taught the
integrator to do the same with the velocity vector it uses in the
Runge-Kutta process.

Anyway, your approach seems safer. I think the modification should be
done right before inserting the freshly calculated point in the structure
outputPoints (in the vtkStreamTracer.cxx that I downloaded on
Tuesday, this happend in lines 883-886).

However, there are two problems:
- shifting the stream line generation to the adjacent face, when the 
stream line
reaches an edge;
- effectively reaching the edge, because the line that is generated 
almost never
touches the edge: the last step in the integration process computes a 
point that is
outside the cell boundaries, and this point is not added; the 
last-but-one computed
point is almost never on the edge.

Prolongating the streamline over adjacent cells is important for the problem
I'm trying to visualize. It remains, certainly, to define a strategy for 
ending the
streamline generation process, but that's a problem for later.

Petru
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.vtk.org/pipermail/vtkusers/attachments/20040730/b9f5b83f/attachment.htm>