Open cmeyer opened 3 months ago
What would be the default display type for shifts of an length N sequence of 2D images, which is N x 2?
Whatever behavior happens now would be the default (displays as image). In addition, in the short term, I would propose adding a special "redimension" processing: "redimension to stack of 2 spectra of length N", similar to how one can redimension it now to be "redimension to sequence of. spectra of length N", which would display as a line plot by default.
Overall, I would also propose the following:
If you want different behavior, let me know. I think the key item is the last one (vector valued arrays), but also the most difficult.
Shifts for sequences of 2D images are a special case that deserves special handling. Currently Andreas experimental align image sequence automatically sets the display type for the shifted output to be a line plot. That has been a huge quality of life improvement to the previous behavior where the shifts were shown as an image, and I would be unhappy if that behavior went away.
When I do need to use the multidimensional processing menu measure shifts, that shift data is shown as an image, and then I used the display type dropdown in the inspector to change it to a line plot display. I would prefer if it was automatically shown as a line plot, but at the very least I don't want to lose an easily available option to display it that way. A menu item to redimension the data sounds like a downgrade from the current workflow.
Good point. I added the issue below and made this issue dependent on it.
The display type pop-up in the inspector is redundant. The display type should be determined entirely by the number of datum dimensions of the data. There is one use case which isn't covered: when the user has something like a width=1024, height=16 image where the 1024 dimension is EELS. In this case, changing the display type was used to display the image as a composite line plot. However, this could be accomplished using 'Redimension' to change the image to a sequence of 16 spectrum of 1024. Needs confirmation that this workflow works as a replacement.