Image capture selection moves out of sight

.

I gather other people have seen this too, but FWIW, I cannot reproduce, at least not with 3.5 alpha. Can someone post more precise steps?

I can't reproduce mine, either (report removed). I "where the #@ is the box?"ed many times, but never tried toggling it. It seems to retrieve the box adequately.

On the other hand I see other glitches, some that have been around, some maybe new with 3.5. If I enable then disable the image capture, the Inspector still reads "Lasso" rather than "Nothing selected", similar for status bar. That much was true in 3.4.2, but for 3.5 builds, I'm also sometimes finding I can't successfully select anything after that. If I click something else, sometimes it works normally, but other times the Inspector updates but I get no actual indication in the score that the selection happened, and some operations that depend on a selection act like there is nothing selected. Unfortuantely, this is quite sporadic.

Looking at the code, I can see we never clear the "edit element" after leaving fotomode, but simply adding a line to do that doesn't change the fact that the Inspector & status bar still have the old info.

Anyhow, there's definitely something not right here, but I'm having trouble pinning it down.

I see. I believe the bug only appears you move the region between pages without first closing it. FWIW, closing and reopening it is usually the easier way of doing things in these cases anyhow, faster than actually dragging the region between pages. If you close it, then simply scroll to the new page, then reopen the image capture, it automatically shows up on the current page. As far as I can tell that's more efficient, anyhow, plus it has the advantage of working :-)

"Neglecting to close it" is extremely common (by me, at least), because the usual action after raising it is copying it and pasting into email. My next action is often to navigate (scroll) to some other place and try to raise the camera window there.

Yes, but again, this works fine - just close and reopen the capture window. Whether you do it before scrolling or after doesn't matter. As far as I can tell, it's the drag from page to page that triggers the bug. And that's actually kind of painfully difficult to do, especially if you're scrolling more than just to the next page, or aren't zoomed out far enough to see two pages at once. So, after scrolling to the new location, instead of awkwardly trying to grab the region from its old location, just click twice on the camera icon. Or use a shortcut if you've defined one.

Good point! Well, in the right-click menu you will also see commands to save up to four different sizes of region, so you can easily set a specific size. So, get it the way you like it, the use Set Standard Size to save it as one of the four presets (for example, Set size A). Then next time, just use Resize to A.

An extra step so not as good a workaround for these cases, but still probably better than most others.

BTW, though, if you are literally using the same size all the time, you could consider using page breaks and appropriate page sizes so it's always exactly a full page. Then you could use the "Auto-resize to page". Or you could just export all pages at once with File / Export. I believe elsewhere maybe you've mentioned doing this. The best solution depends on the specifics of your situation.

Meanwhile, it seems the bug happens upon moving the frame. I looked at the code for handling that but it's kind of Greek to me. Maybe someone who understands more about transforms and matrices and so forth would want to look at this...

I can reproduce, always and easily (no need to edit or save the image), on 3.4.2 and 3.5 Alpha (Windows 10)
Steps:
1) SATB + Organ template : satb.mscz
2) Enable Image Capture tool
3) Drag the selection to the second page
4) Drag to the third page and Maj + right cursor key

Result: selection disappears

Well, after further tests, the selection don't exactly disappears. It is located (it jumped) one or two pages further, astride two pages, and of a much larger size - image below
See, same score, after appending 100 measures (and system breaks every 4 measures) : satb +.mscz
I see more or less the same behaviour since the version 3.0.0
I just remark a change in February 2018 (the size of the default selection has been significantly reduced) . But I don't know if it's causally related. Not sure.

regression vs. what earlier version?

Works as expected in version 2.3.2.

I see

Came up again in https://musescore.org/en/node/315025

Can verify it's happening in 3.6.2 as well.