BigWarp - impossible to remove landmark?

Image Analysis
,
#1

Hello everybody and in particular @bogovicj,

I really find that BigWarp is super useful, but I have a hard time with the landmark placement. I have two questions:

  • What’s the logic behind pairing multiple points between the fixed and moving images ? So let’s say I put four points in the fixed image (pf1, pf2, pf3, pf4) then 4 points in the moving image (pm1, pm2, pm3, pm4). To me the logical thing would be that the resulting pairing is pf1->pm1, pf2->pm2,pf3->pm3,pf4->pm4. But it’s not. The current result is pf1->pm2, pf2->pm3,pf3->pm4,pf4->pm1. So at the end, it means that I can only put one point after the other (it’s a bit tedious). Is there a way to do better ?
  • I really couldn’t find a way to delete landmarks (I can inactive, but can’t delete). However in the documentation it’s written that ‘right-click + delete’ in the landmark table should delete the landmark (https://imagej.net/BigWarp#Notes_on_point_addition_and_landmark_pair_selection). But for me it doesn’t do anything. Is this a bug ? Or has this feature been replaced by something else ?

Thanks for the tips,

Nicolas

PS: I think I’ve found the answer to question 1, but I’ll just let it for the record

1 Like
#2

Hi @NicoKiaru !

Glad you find it useful.

On the second point (not removing landmarks):

On the second point (not removing landmarks):
That is indeed a bug, sorry about that. It should be fixed in the latest build downloadable from here.

I plan to make a new release this month - there are a few lingering issues I’d like to deal with before release time.

On the first point (order of landmark placement):

“short answer”

Agreed that its tedius.

The “better” (but non-intuitive from one perspective) way is to add points in one window:
pt1 > pt2 > pt3 > pt4
Then do add points in the second window:
pt4 > pt1 > pt2 > pt3

Or in the first window:
pt1 > pt2 > pt3 > pt4
Then click pt1 in the table.
Then in the second window:
pt1 > pt2 > pt3 > pt4

Short summary of the behavior:

  • When you click, you affect “selected” points
  • The last point you clicked is “selected”

“long answer”

I’ll explain what the logic for the above is and let me know if this is consistent with what you find.

The behavior:

  • If you don’t have any points selected, or if the selected row is “complete” (i.e. has both moving and fixed points), and you click in either window - add a new point
  • If you click a “new” (unpaired) point, the point you just added is selected
  • If you click in the moving (fixed) window, and the selected point already has a moving (fixed) point, then add a new point.
  • If you click in the moving (fixed) window, and the currently selected point is missing a moving (fixed) point - complete the row
    • If there are “incomplete rows”, select the next incomplete row (smallest indexed row greater than the just completed one, wrapping to zero if necessary).

I know its not intuitive at first glance, and I really don’t like that about it…but it’s at least unambiguous. I very well could be falling into the overthinking trap, and making the common case hard and the rare cases easy.

Thanks for the feedback, I hope this helps,
John

1 Like