- Open MITK-Data/3D+t-Heart
- Switch on volume visualization 💣
- Crash 💥
Description
Revisions and Commits
rMITK MITK | |||
Restricted Differential Revision | rMITKf68ef917dff3 Fix volume rendering for 4d images, adapt histogram and volume itself | ||
Restricted Differential Revision | rMITKed883fa6af57 Refactorings from review | ||
Restricted Differential Revision | rMITKfd647480c6ec Fix volume rendering for 4d images |
Related Objects
- Mentioned In
- T28902: 2021 Week 48 (Very Early December)
T28274: [Volume Visualization] Volume Rendering for 4D images does only work on timestep 0 or not at all - Mentioned Here
- T28274: [Volume Visualization] Volume Rendering for 4D images does only work on timestep 0 or not at all
T24766: Last time step of 3D+t data with arbitrary time geometry not selectable by image navigator spinbox
T24767: Workbench crashes when loading two 3D+t images with different number of time steps
Event Timeline
Segmentation fault occurs in vtkAOSDataArrayTemplate::GetTypedComponent(vtkIdType tupleIdx, int comp) after accessing the buffer[0].
Crash only occurs with the DICOM 3D+t image which also causes trouble in T24767 and T24766. Seems to work with all other 3d+t images in MITKData. Volume visualization works for the image if it is exported to .nrrd first.
We found out that volume visualization works as expected when following the described steps:
- Close rendering display
- Open MITK-Data/3D+t-Heart (rendering display is opened automatically)
- Switch on volume visualization
- No crash :)
We couldn't find out the difference between the initial rendering display and the second rendering display so far.
The issue still occurs on both Windows and Linux systems. The transfer function widget seems to be initialized correctly. The last MITK method called before diving into VTK internals mentioned above is VtkMapper::MitkRenderVolumetricGeometry.
I remove this task from the beta backlog. If there's a strong interest in having this fixed in the upcoming release, someone has to fix it this week and assign the MITK (2018-04) tag again.
Cannot reproduce this behavior with the current master branch anymore. Now it always crashes.
This issue still occurs. Following previous finding still holds true: Volume visualization works for the image if it is exported to nrrd first and then loaded as nrrd.
Seems that the pixel type is not set properly:
Show details lists "Channel: Unnamed [scalar], PixelType: scalar" for the nrrd file and "Channel: Unnamed [], PixelType: " for the DICOM file.
Could this be the reason for the access violation later on?
I couldn't reproduce this bug in the latest Release installer on Windows. However, there is no volume visualization shown at all (no matter which timestep or which histogram I select).
If the crash does not occur anymore on Ubuntu etc. we can resolve the crash bug.
I mentioned the general problem with 4D data and visualization here: T28274