[vtk-developers] How do I reliably the origin of a vtkImageData in VTK 6?

[Cory Quammen]

David,

Thanks for the quick reply. I'm relieved to hear that the changes in
VTK 6 haven't introduced any kind of odd synchronization issues
between what the pipeline believes about a vtkImageData and what the
vtkImageData believes about itself.

This looks like an easy bug fix. I'll post a patch to gerrit shortly.

Cory

On Thu, Oct 11, 2012 at 10:46 AM, David Gobbi <david.gobbi at gmail.com> wrote:
> Hi Cory,
>
> After you call extractor->Update(), it will have updated the whole
> extent and nothing that occurs downstream should cause any changes to
> the extractor's output.  None of the VTK 6 changes should have changed
> this behavior.
>
> I've looked through the code in both vtkExtractVOI and in
> vtkMetaImageWriter.  The code in vtkExtractVOI looks fine.  But in
> vtkMetaImageWriter I found this in the Write() method, around lines
> 114 and 148:
>
>   double * origin = this->GetInput()->GetOrigin();
>   ...
>
>   origin[0] += ext[0] * spacing[0];
>   origin[1] += ext[2] * spacing[1];
>   origin[2] += ext[4] * spacing[2];
>
> The writer is modifying the input's origin.  A "git blame" shows that
> this bug was introduced in 2007, so it's been there for a long time.
> The bug manifests itself any time that someone writes a metafile where
> the Extent of the input doesn't start at zero.
>
>  - David
>
>
> On Thu, Oct 11, 2012 at 8:11 AM, Cory Quammen <cquammen at cs.unc.edu> wrote:
>> [Reposting to the VTK developers list]
>>
>> I'm using VTK at commit 5754065bad83bf9cbcfbcfa72ecfbbed91f277db (July
>> 13, 2012).
>>
>> Goal:
>>
>> I'm attempting to something with a vtkImageData object outside a VTK
>> pipeline that I think should be straightforward (rasterizing a
>> sphere). To do this properly, I need to know the origin and spacing of
>> the vtkImageData object produced from a filter.
>>
>> Problem observed:
>>
>> When I ask a vtkImageData about it's origin, the answer I get depends
>> on whether it has been processed by a downstream filter or not. Here's
>> what I'm seeing.
>>
>> I have a short pipeline sketched here:
>>
>> vtkExtractVOI * extractor = vtkExtractVOI::New();
>> extractor->SetInputData( image );
>> extractor->SetVOI( voi ); // Assume voi[0] > image->GetExtent()[0]
>> extractor->Update();
>> // extractor->UpdateWholeExtent(); // using this makes no difference
>>
>> vtkImageData * subImage = extractor->GetOutput();
>>
>> // Ask subImage for it's origin
>> double originBeforeWrite[3];
>> subImage->GetOrigin( originBeforeWrite );
>>
>> // Save the extracted region to a file
>> vtkMetaImageWriter * writer = vtkMetaImageWriter::New();
>> writer->SetFileName( ... );
>> writer->SetRAWFileName( ... );
>> writer->SetInputData( subImage );
>> writer->Write();
>>
>> // Ask subImage for it's origin again
>> double originAfterWrite[3];
>> subImage->GetOrigin( originAfterWrite );
>>
>> Now, when this code is executed, originAfterWrite[i] != originBeforeWrite[i].
>>
>> This is disturbing.
>>
>> I'm guessing writing subImage to a file triggers an update to the
>> origin in the pipeline that then gets set back in subImage, and that
>> explains why originAfterWrite[i] != originBeforeWrite[i]. Shouldn't
>> the origin returned from subImage->GetOrigin() be updated when I
>> invoke extractor->Update()?
>>
>> If not, could you please advise me on the proper way to get the origin
>> from a vtkImageData that is produced as the output of a filter? I
>> looked in the VTK 6 migration guide, but couldn't find anything
>> pertaining to this question.
>>
>> Thank you,
>> Cory
>>
>> --
>> Cory Quammen
>> Research Associate
>> Department of Computer Science
>> The University of North Carolina at Chapel Hill
>> _______________________________________________
>> Powered by www.kitware.com
>>
>> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
>>
>> Follow this link to subscribe/unsubscribe:
>> http://www.vtk.org/mailman/listinfo/vtk-developers
>>



-- 
Cory Quammen
Research Associate
Department of Computer Science
The University of North Carolina at Chapel Hill