[vtkusers] Memory problem

[Raúl Ferriz]

El 15/07/2010 18:24, David Gobbi escribió:
> On Thu, Jul 15, 2010 at 9:56 AM, Raúl Ferriz<raul at torresyvalero.com>  wrote:
>    
>> Hi David!
>>
>> My pipeline looks like:
>> vtkGDCMImageReader -->  vtkImageChangeInformation (to correctly set ZSpacing)
>> Now from this output
>> ->vtkImageShrink3D->vtkContourFilter->vtkPolyDataMapper->vtkActor
>> and also the output from vtkImageChangeInformation goes to 3 vtkImageReader2
>> (one for each slicer)
>>
>> Is there a way to optimize this?
>>      
> I made a mistake, when you said "axial, sagittal, coronal", I did not
> thing you were doing contouring.  I thought you were extracting
> slices.
>
> For your pipeline, the only improvement is maybe to use just one
> vtkImageShrink3D instead of using one per view.  You don not have to
> worry about vtkImageChangeInformation using memory, because it is a
> special filter that doesn't use any memory (it just passes the data
> while changing the information).
>
>     David
>
>    
Sorry if I did not explained correctly:
I only have one 
vtkImageShrink3D->vtkContourFilter->vtkPolyDataMapper->vtkActor
This pipeline is to show a 3d view of dicom on one renderwindow.

But I have 3 vtkImageReader2 to show every plane (axial, sagittal, 
coronal) on his own renderwindow.

Well, then my only solution is to change my platform ? There is not any 
way to force vtk to try split memory allocation when malloc fails? Or 
force to use swap when malloc fail?

Many thanks for your interest