[Paraview] Inconsistent results with the stream trace filter

[Yuanxin Liu]

Hi, Joe,
  Berk reminded me that Andy Bauer has put in some fixes for gradient
interpolation for pyramids (5a7dc931c54d868).  Since your data contain
pyramids, this is quite likely the fix for this bug.
Leo

On Fri, Oct 19, 2012 at 12:05 PM, Yuanxin Liu <leo.liu at kitware.com> wrote:

> Hi, Joe,
>   It looks like this was indeed a bug in 3.14.1 but the bug has been fixed
> in the current development branch. I am not entirely sure which commit
> explicitly did it.
>
> Leo
>
> On Fri, Oct 19, 2012 at 5:54 AM, Joe Borġ <mail at jdborg.com> wrote:
>
>> Hi,
>>
>> I am using 3.14.1 64-bit on a single core (if I enable multi-core, it
>> becomes very unstable).
>>
>> Adding tets to this small case works, but I've got a bigger duct with a
>> more refined mesh and adding tets don't fix the problem (make it better
>> though).
>>
>> See attached PVSM.
>>
>>
>>
>> Regards,
>> Joseph David Borġ
>> http://www.jdborg.com
>>
>>
>> On 18 October 2012 19:00, Yuanxin Liu <leo.liu at kitware.com> wrote:
>>
>>> No, I did not use Tetrahedralize.
>>>
>>> Are you using the lastest development version?  Are you running paraview
>>> as a single process?
>>>
>>> Do you mind saving a state file of your session and post it?
>>>
>>> thanks!
>>>
>>> Leo
>>>
>>>
>>>
>>> On Thu, Oct 18, 2012 at 4:31 AM, Joe Borġ <mail at jdborg.com> wrote:
>>>
>>>> Hi Leo,
>>>>
>>>> You don't seem to have used the Tetrahedralise filter?  How have you
>>>> got this to work without that?
>>>>
>>>> Mine works fine after setting the fluid as input and source as the
>>>> inlet, but I must first use the Tetrahedralise filter.
>>>>
>>>>
>>>> Regards,
>>>> Joseph David Borġ
>>>> http://www.jdborg.com
>>>>
>>>>
>>>> On 17 October 2012 18:14, Yuanxin Liu <leo.liu at kitware.com> wrote:
>>>>
>>>>> Hi Joe,
>>>>>
>>>>> I am not seeing exactly what you are seeing.  In the attached image
>>>>> produced with the current master, I created a stream tracer with outlet
>>>>> block as the seeds and the fluid block as the input, and the streamlines
>>>>> seem right.
>>>>>
>>>>> But I think there is a bug in how the stream tracer handles multiblock
>>>>> data, because if I change the stream tracer input to the whole pipe data
>>>>> set, instead of just the fluid block, some stream lines go straight to the
>>>>> void like you described.
>>>>>
>>>>> Can you try setting up the pipeline as I did using ExtractBlock and
>>>>> compare with my image?
>>>>>
>>>>> Leo
>>>>>
>>>>> On Wed, Oct 17, 2012 at 10:51 AM, Magician <f_magician at mac.com> wrote:
>>>>>
>>>>>> Hi Joe,
>>>>>>
>>>>>>
>>>>>> "Tetrahedralize" filter is applicable.
>>>>>> All cells are splitted into Tetrahedral ones.
>>>>>> Try it.
>>>>>>
>>>>>>
>>>>>> Magician
>>>>>>
>>>>>>
>>>>>> On 2012/10/17, at 23:40, Joe Borġ wrote:
>>>>>>
>>>>>> > Thanks Magician,
>>>>>> >
>>>>>> > Thanks for testing.  What are you using to convert to tet?  Is it a
>>>>>> ParaView feature?
>>>>>> >
>>>>>> > Leo, can the priority of the bug be raised please?
>>>>>> >
>>>>>> >
>>>>>> > Regards,
>>>>>> > Joseph David Borġ
>>>>>> > http://www.jdborg.com
>>>>>> >
>>>>>> >
>>>>>> > On 17 October 2012 14:20, Magician <f_magician at mac.com> wrote:
>>>>>> > Hi Joseph,
>>>>>> >
>>>>>> >
>>>>>> > I tried your data and got a result as yours.
>>>>>> > Your data only includes Hex, Wedge, Tet and Pyramid,
>>>>>> > but streamlines are stopped earlier.
>>>>>> >
>>>>>> > Same as my polyhedral pipe sample, I applied Tetrahedralize filter
>>>>>> > to your pipe source, and I could get valid streamlines.
>>>>>> >
>>>>>> >
>>>>>> > Magician
>>>>>> >
>>>>>> >
>>>>>> > > Pipe.tar.gz attached to the bug report below, in the form of an
>>>>>> Ensight
>>>>>> > > Case.  First extract the archive and to open in Paraview, use
>>>>>> "All files
>>>>>> > > (*)" when browsing and select the .encas, then select "EnSight
>>>>>> Files".
>>>>>> > >
>>>>>> > >
>>>>>> > > Regards,
>>>>>> > > Joseph David Borġ
>>>>>> > > http://www.jdborg.com
>>>>>> > >
>>>>>> > >
>>>>>> > > On 12 October 2012 03:28, Yuanxin Liu <leo.liu at kitware.com>
>>>>>> wrote:
>>>>>> > >
>>>>>> > > > Hi,
>>>>>> > > >    Is it possible for you to upload the data?  By the way, if
>>>>>> the cell
>>>>>> > > > type is polyhedron, then this is quite likely the same bug as
>>>>>> this:
>>>>>> > > > http://paraview.org/Bug/view.php?id=13442
>>>>>> > > >
>>>>>> > > > Leo
>>>>>> > > >
>>>>>> > > > On Thu, Oct 11, 2012 at 6:17 AM, Joe Borġ <mail at jdborg.com>
>>>>>> wrote:
>>>>>> > > >
>>>>>> > > >> Hi guys,
>>>>>> > > >>
>>>>>> > > >> I've got an odd issue with the stream trace and stream trace
>>>>>> custom
>>>>>> > > >> source filters in Paraview (3.14.1 64bit).  No matter what
>>>>>> settings I use,
>>>>>> > > >> the traces die off very soon and some even carry on straight
>>>>>> (even when
>>>>>> > > >> there's no fluid).
>>>>>> > > >>
>>>>>> > > >> See the two screenshots I've got from Paraview.
>>>>>> > > >>
>>>>>> > > >> http://jdborg.com/images/paraview/pv_inlet.png
>>>>>> > > >> http://jdborg.com/images/paraview/pv_outlet.png
>>>>>> > > >>
>>>>>> > > >>
>>>>>> > > >> Not sure if this is a setting that's easy to miss or if it's
>>>>>> more
>>>>>> > > >> terminal.
>>>>>> > > >>
>>>>>> > > >> Thanks,
>>>>>> > > >>
>>>>>> > > >>
>>>>>> > > >> Regards,
>>>>>> > > >> Joseph David Borġ
>>>>>> > > >> http://www.jdborg.com
>>>>>> >
>>>>>>
>>>>>>
>>>>>
>>>>
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.paraview.org/pipermail/paraview/attachments/20121019/9cc3b965/attachment.htm>