Usability issues, feature requests #152
Comments
|
Hi @acardona I hope Bigwarp has been useful, and thanks for the feedback! you have lots of good points.
bigdataviewers bookmarks sort-of do this, but structuring them better would be a great feature, agreed.
There are many ways to navigate to landmarks
(some of those need documenting - whoops)
This is possible with
Increasing the landmark size will make them easier to select.
and
auto-synching is a good idea, but this is easy to do yourself with my favorite hotkeys:
And as you said, many of your other ideas are "bdv" ideas, but I agree, those would be nice to have.
I'm going to be pedantic here. The transformation moved the landmark away from the view, but can see how that would feel bad. Some tips or "how I like to work with bigwarp" Another nice pattern if your images are close is to shift-click + drag when you have the moving image displayed. Usually I'll do a smallish number of landmarks to get close, then find nearby things, use shift+click+drag in one plane, then change the plane to check / adjust the other orientations Thanks again! [1] because once a pair is completed, it triggers an update of the transform. |
When using the bigwarp over several hours to painstakingly align an LSM to an EM volume, a number of usability issues became evident:
A proposed solution: store a sequence of views, each one stored upon mouse or key release. To reduce the listing of keys, compress multiple subsequent key presses (key presses within less than a second from each other) for the same key into just one event. Then offer ways to navigate these past views.
Another, complementary solution: the "Save settings" must also store the views of each volume.
Yet another complementary solution: like in CATMAID, add means to store named views. So that one can swiftly go back to that view, by choosing it from a menu. And have these stored views be saved in the settings XML too.
Yet another: allow sort of locking the view, or rather constraining the possible movements, so that only panning and view Z axis movement is allowed. This restriction alone would do wonders.
The above are general issues with the BigDataViewer.
Upon adding a new landmark in fused transformed mode, the transformation moves the view away from the landmark. Automatically scroll in the Z axis (relative to view) to find it again.
When in transformation mode, offer means to have the same view on both the moving and fixed image. As in, add a menu item in the fixed image window to see it in the same view than the moving image, and viceversa. This alone would cut down user time enormously.
Selecting landmarks is too difficult. The '+' pointer is as large. Make perhaps one like in CATMAID: an empty circle, so one can see what is inside. Additionally, the landmark ought to grow in diameter, or change color, or something, to notify the user that the mouse has gotten close enough that it can click on it to select it, instead of adding a new landmark.
Finding landmarks is too difficult. I would like to scroll while e.g., holding down alt key, or something, and have the scrolling in the view Z axis stop when it hits a landmark. Or draw a rectangle and have the view automatically scroll down or up to find the nearest landmark. Or some such system.
Consider enabling changing the diameter of landmarks and their colors, so they are more visible.
Enable/disable synchronizing moving and fixed windows when in transformation mode. So that one can navigate both simultaneously without having to guess.
When exiting transformation mode, the moving image is often terribly zoomed out or zoomed in, and often outside the field of view. One has to go to the landmarks, double-click on one to bring the volume into view, then zoom in or out. Make this easier by storing the view one had prior to entering transformation mode, and then restoring the view to that. Here, solutions to (1) above would also help lots.
When in transformation, fused mode, moving a landmark is often catastrophic: a very strong, warped deformation. Control+z works here, thankfully, but why should this happen in the first place. A toggle to prevent dragging landmarks would be helpful.
Another BigDataViewer issue: show me info. As in, which images are being shown? If the image is wrapping an ImagePlus, tell me about the image properties of that ImagePlus. In particular the calibration and the file path in the hard drives alone would be very helpful.
Related, and BigDataViewer issue: have a button that emits a script to access the underlying data. As in, via Frame.getFrames(), traverse down the swing or awt components hierarchy until reaching the class that points to the data. Emiting this 2-3 line script would enable one to work on the data that BigWarp is using. To change its values, to measure it, to visualize it elsewhere simultaneously.
Another related to BigDataViewer: enable browsing without pixelation. As in, the data is loaded already, no need to use first the low-res levels of the multi-res pyramid. Use directly the best level, and render directly with interpolation if that's enabled. CPUs are fast and can deliver seamless browsing anyway. The pixelation can be extremely distracting.
Thanks so much for BigWarp!
The text was updated successfully, but these errors were encountered: