[Paraview] Trouble making stream traces

[Kent Eschenberg]

Last year I worked with a dataset that had similar problems but with a 
slightly different cause. It turned out that many vertices that should 
have been shared between adjacent cells were actually two vertices close 
together. As a result the very thin, almost invisible gap between the 
cells halted the streamline algorithm (it thought it had reached the end 
of the dataset). One of the filters (I think it was clean-to-grid) took 
care of most of this but eventually we had to fix the application that 
generated the mesh.

Kent
Pittsburgh Supercomputing Center

Dan Goldstein wrote:
> Hi John,
>   I too have found the stream traces filter hard. While I have not 
> figured it all out it seems most of the troubles I have had were related 
> to bad cell connectivity in the data set. I took your data and made a 
> little picture with stream traces. To get it to work I had to first run 
> through the "tetrahedralize" filter. I believe that doing this fixes the 
> internal cell connectivities. I am sure someone from KiteWare can give a 
> real answer to what it is really doing. I hope this helps:
> 
> Dan
> 
> John Fettig wrote:
>> I'm having trouble creating stream traces on a data set that I
>> generated...it seems no matter what parameters I choose, I can't get
>> more than a fraction of the domain swept.  Could somebody take a look at
>> my dataset and see if there is something obviously wrong with it (I
>> generate the vtu XML file myself, from inside a fortran code), or how I
>> can get stream traces?
>>
>> http://mithlond.mechse.uiuc.edu/~jfettig/velo0000000000.vtu.bz2
>>
>> John