Unexpected behavior of `edu.stanford.rsl.conrad.utils.ImageUtil.wrapGrid3D` about conrad HOT 3 CLOSED

As far as I can see, this is only called for MultiChannelGrids. Those are currently only used in material decomposition, multi-energy processing, and image segmentation. As the ImagePlus container is only used to write data to disk or display, I would expect that changing this to the current grid is OK and does not cause any trouble.

However, you need an adopted version anyway as the multichannel data is just a new channel in the same resolution and not a 3D volume. As such the current behaviour should be incorrect anyway. I guess, it was never noticed as it is typically only used for display with ImageJ. I would support fixing this issue.

from conrad.

See pull request #141

Multi-channel handling
MultiChannelGrids were indeed not recognized as such.
MultiChannelGrid3D supports multple channels by using an internal Grid4D data member. As such checking the subgrids for instanceof MultiChannelGrid2D was incorrect.
The method now checks the Grid3D directly and extracts information accordingly.

Calibration
The version of setCalibrationToImagePlus without an input grid was removed. Both default and multi-channel grids carry the properties needed to fill the ImagePlus calibration. Consequently, checking the global configuration for this information should not be necessary at all.

from conrad.

There are in fact some occasions (e.g. in tutorial.motion.MotionCompensatedReconExample), where Grid3D was used as a container for MultiChannelGrid2D without being multi-channel itself. This required the multi-channel property to be checked on the subGrids. Hence, we re-added (and reworked) this part.
Nevertheless, in case of the Grid3D container, calibration can still be acquired from the 2D subgrids directly using setCalibrationToImagePlus2D. setCalibrationToImagePlus without an input grid remains removed.

With the merge of pull request #141, I'd support closing this issue.

from conrad.