[Paraview] Recent Xdmf volume rendering

Dominik Szczerba dominik at itis.ethz.ch
Wed Dec 2 09:31:11 EST 2009


Yes this is today's CVS, compiled in RELEASE, so no backtrace available.
It does not crash, just invalidates the information upstream.

I do not know when this started happening because it is only today that 
the ExtractSubset at all started working for me.

To reproduce:

1) load 3D transient XMF Uniform 3DCoRectMesh dataset
2) chose ExtractSubset, select a shrinked VOI.
3) (what I do now is volume rendering of the fragment)
4) manipulate the VOI in the ExtractSubset filter a few times.
5) after a few changes the upstream information is damaged (i.e. the 
current VOI has overwriten the original extents and not whole dataset is 
available to subsetting anylonger).

To deliver an example dataset will not be easy (mine is big, 
confidential etc.) but if you cannot reproduce this problem using some 
other test datasets I will try fabricate one for you.

Dominik

Berk Geveci wrote:
> Hmm. This is not good. I am guessing this is in CVS ParaView? Do you
> know when this started happening? Any suggestions on how to reproduce
> it? Also, can you run in a debugger and give us stack traces?
> 
> On Wed, Dec 2, 2009 at 7:51 AM, Dominik Szczerba <dominik at itis.ethz.ch> wrote:
>> However: changing ExtractSubset parameters (output extents) at some point
>> (after a few changes) "propagates" upstream, i.e., the original dataset
>> "inherits" the extents of the downstream filter (gets truncated). This is
>> clearly wrong, but probably unrelated to this topic.
>>
>> BTW. For a longer while I have been observing dramatic drop in PV usability.
>> Average time before crashes is only a few minutes. I am basically scared to
>> touch anything once I set up my pipeline (change names of calculator
>> variables, extents of VOI etc. etc.). My last resort of using save states is
>> now failing too, it is usually impossible to reload the state (segmentation
>> faults).
>>
>> Dominik
>>
>> Dominik Szczerba wrote:
>>> I checked the latest CVS (compilation errors see the other thread) and the
>>> problem is gone, ExtractSubset behaves now correctly.
>>>
>>> Thanks
>>> Dominik
>>>
>>>
>>>
>>> Utkarsh Ayachit wrote:
>>>> Possibly related. Any possibility you can share a concrete
>>>> example/dataset?
>>>>
>>>> Utkarsh
>>>>
>>>> On Tue, Dec 1, 2009 at 4:49 PM, Dominik Szczerba <dominik at itis.ethz.ch>
>>>> wrote:
>>>>> Just on the occasion of transient 3DCoRectMesh datasets being loaded as
>>>>> "Image (Uniform Rectilinear - I just noticed that "Extract Subset" goes
>>>>> completely crazy - the output is wrongly sized and contains no cells...
>>>>> Accidental or related?
>>>>>
>>>>> Dominik
>>>>>
>>>>>
>>>>>>>>> Grid) with blanking"
>>>>> Eric E. Monson wrote:
>>>>>> That seems to work great now, Utkarsh -- Thanks for the fast fix!
>>>>>>
>>>>>> Can someone please explain what the "with blanking" is, though -- Is it
>>>>>> a
>>>>>> way to mask out certain blocks, or something? Is this something that is
>>>>>> supported within Xdmf? Not a huge issue, but I was just curious...
>>>>>>
>>>>>> Thanks again,
>>>>>> -Eric
>>>>>>
>>>>>>
>>>>>> On Nov 26, 2009, at 12:31 PM, Utkarsh Ayachit wrote:
>>>>>>
>>>>>>> There was another bug in the code that was incorrectly publishing
>>>>>>> origin of the dataset as spacing which could have been messing up the
>>>>>>> volume rendering. I've committed a fix. Feel free to give it a try
>>>>>>> now.
>>>>>>>
>>>>>>> Utkarsh
>>>>>>>
>>>>>>> On Thu, Nov 26, 2009 at 12:07 PM, Utkarsh Ayachit
>>>>>>> <utkarsh.ayachit at kitware.com> wrote:
>>>>>>>> To keep the behavior consistent with the previous reader, I've
>>>>>>>> changed
>>>>>>>> the vtkXdmfReader2 to produce vtkImageData for 2D/3DCoRectMesh.
>>>>>>>>
>>>>>>>> Utkarsh
>>>>>>>>
>>>>>>>> On Wed, Nov 25, 2009 at 4:33 PM, Eric E. Monson <emonson at cs.duke.edu>
>>>>>>>> wrote:
>>>>>>>>> Hey all,
>>>>>>>>>
>>>>>>>>> I just noticed today that I can't volume render my Xdmf ImageData
>>>>>>>>> sets
>>>>>>>>> in
>>>>>>>>> CVS ParaView (today's -- OS X 10.5.8, serial build, Qt 4.5.2 cocoa,
>>>>>>>>> cmake
>>>>>>>>> 2.9.20091003).
>>>>>>>>>
>>>>>>>>> With the binary PV 3.6.1 I can load in a temporal collection or
>>>>>>>>> single
>>>>>>>>> 3DCoRectMesh and it  volume renders fine. With current 3.7 (now
>>>>>>>>> using
>>>>>>>>> vtkXdmfReader2, I assume), this data shows up as "Image (Uniform
>>>>>>>>> Rectilinear
>>>>>>>>> Grid) with blanking" and I can choose volume rendering but nothing
>>>>>>>>> shows up
>>>>>>>>> in the render window.
>>>>>>>>>
>>>>>>>>> Volume rendering of a Wavelet Source works just fine, so I'm
>>>>>>>>> wondering
>>>>>>>>> what
>>>>>>>>> the "with blanking" part is, and if that's what's causing my
>>>>>>>>> trouble?
>>>>>>>>>
>>>>>>>>> Thanks,
>>>>>>>>> -Eric
>>>>>>>>>
>>>>>>>>> ------------------------------------------------------
>>>>>>>>> Eric E Monson
>>>>>>>>> Duke Visualization Technology Group
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> _______________________________________________
>>>>>>>>> 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
>>>>>>>>>
>>>>>> _______________________________________________
>>>>>> 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
>>>>>>
>>> _______________________________________________
>>> 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
>>>
>> _______________________________________________
>> 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
>>
> 



More information about the ParaView mailing list