I do not think that this is a rendering problem.  I think you have just fooled 
the filter that extracts the external faces of the blocks (which is what is 
actually rendered).  Notice that when you arrange the overlapping meshes 
exactly as you are doing, that some of the faces are coincident.  In 
particular, the face that is missing is coincident.  Up to the point before you 
run clean to grid, these faces are topologically distinct because they use an 
independent (although coincident) set of points.  When you run clean to grid, 
these coincident points are merged.  When the points are merged, the cells from 
the two meshes now share the same indexed points.  At this point, these faces 
are no longer topologically distinct.  The external faces filter notices that 
these two faces use the same points and marks it as a neighbor face (this is 
the definition of a neighbor face) and "correctly" removes it.

The problem is basically that when you append the data sets you are creating a 
non-manifold mesh.  To correct the problem, you need some sort of filter that 
identifies coincident cells and removes duplicates.  I know of no such filter, 
so you will probably have to write your own.

-Ken


On 11/11/09 7:33 AM, "Sven Buijssen" <[email protected]> wrote:

Hi,

I've come across something that I consider a rendering bug, but before filing a
bug report I'd like to discuss it first. Maybe it sounds familiar. (I'm guess
it's not related to the z-fighting issue discussed last week.)

I have two volume grids consisting of hexaeder which partially overlap (wavelet
sources for demonstration purposes in the attached *.pvsm file). After applying
AppendDataSets and CleantoGrid filter it looks like the resulting grid contains
all points and cells except the cell at the intersection of the two grids. See
attached image.
But the "missing" cell is actually there. Twice. It can be confirmed by applying
a ProgrammableFilter to add a cell array with cell numbers and applying a
Threshold filter. Set both lower and upper threshold to either 0 or 30 to verify
that the "missing" cell is still there. It simply not rendered any more.

It happens with PV 3.6.1 binary release and PV compiled from CVS HEAD. (The
state file has been created with PV from CVS HEAD.)

Any thoughts?

Thanks
Sven



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

_______________________________________________
Powered by www.kitware.com

Visit other Kitware open-source projects at 
http://www.kitware.com/opensource/opensource.html

Please keep messages on-topic and check the ParaView Wiki at: 
http://paraview.org/Wiki/ParaView

Follow this link to subscribe/unsubscribe:
http://www.paraview.org/mailman/listinfo/paraview

Reply via email to