Closed xulman closed 1 year ago
Maybe this PR could be the opportunity to improve opening a mastodon file when the image data is missing. For now several things could be improved:
[ ] The user gets a convoluted error message in the IJ console. We could give them a nice warning window, likethe ones we use when opening from ImageJ ImagePlus.
[ ] The BDV appears empty of tracks when you open it. The user must dezoom or pan or zoom to make them appear. Maybe because of an incorrect bounding box determination at start?
[ ] Same thing for time: TrackScheme opens with a max timepoint of 9 and cannot zoom out. It does not determine max timepoint from the data:
yes, agree to all
btw, I've revamped the File -> Fix Image Path dialog in mastodon-tomancak repo recently;
on the master in m.-tomancak repo it looks now like this:
In the dummy sub dialog I could maybe pre-load default values from the Graph
, and we could try to open this thing whenever loading of project fails?
what do you think? @tinevez @tpietzsch @maarzt
so, to the latter two points of JY above:
the dummy data is some default Interval
, likely small.. and thus off compated to where the spots actually are, thus the BDV view looks at nothing
similarly to the timpoints range, the dummy data by default creates only 10 timepoint, which is why TS win was "trimmed" (also BDV window should have time slider only for 10 TPs)
while i didn't care too much about the spatial extent of the dummy data (it's empty, why would I even cared to open BDV...), I cared about the temporal limit and that actually motivated me to introduce the "I want dummy data" button in which especially the time span can be adjusted
actually was now tracing how default dummy data is created:
.xml
data pointer, we open it directly herehmm.. it's a puzzle... when the .xml
file is missing, it does not even reach the inspect()
code...
anyway, only wanted to show that the Dummy class default c'tor is misleading, it's hardly ever used
(sometime later...)
and if it reaches the inspect()
, it never fails, and creates the small dummy data
soo, the default dummy is really only 100x100x100, 10 TPs
origin/dev
brachprintStackTrace()
with getMessage()
... no big deal, but looks less messy@tinevez I would say this could close the item 1 "user gets convoluted message", what do you think?
The fix dialog path, in its "dummy button", got option to adjust the dummy data to be just "around the spots" (aka axis aligned bounding box in spatial axes and temporal axis)...
you can find it in tomancak mastodon repo, branch aabb-dummy
...which may close the item 2 & 3 from here
Hello @xulman I think this PR is ready to be merged. Can you briefly tell me of a procedure that would allow me to test it?
Previously there was BasicImgLoader utlized in
DummySpimData
object. But some code around (even outside Mastodon) is casting toViewerImgLoader
.To enable these pieces of code to function with our dummy data, we had to upgrade the dummy....
The implementation features single resolution level pyramid (I hope).