Support loading pyramids as multiscale layers

[psobolewskiPhD]

This is #51 but with the latest changes merged in and conflicts resolved.
I've tested this with:
0.4.18, 0.4.19 -- old napari async implementation

• very poor performance
• lots of black screens

0.5.0, 0.5.5 -- new napari async implementation

• performant with the NAPARI_ASYNC=1 (or setting in Preferences)! some flashing as things are (re)loaded ([async] Figure out flashing/reloading of recent data, perhaps lack of caching? napari/napari#7391) but performance seems on par with napari-ome-zarr (also with ASYNC set on) for a dataset I can access as ome-zarr in s3 or as ndpi in omero.
• I've also not noticed any memory issues are reported in Pyramid images #51 despite extensive zoom, pan, letting the image sit, whatever.
• If you do NAPARI_ASYNC=0 everything still works, you just get blocking of the viewer when zooming requires fetching new data.

Note: #51 dates back to napari 0.4.15

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 62 lines in your changes missing coverage. Please review.

Project coverage is 30.90%. Comparing base (1d53966) to head (a697c85).

Files with missing lines	Patch %	Lines
src/napari_omero/plugins/loaders.py	0.00%	62 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main      #81      +/-   ##
==========================================
- Coverage   33.13%   30.90%   -2.24%     
==========================================
  Files          13       13              
  Lines         845      906      +61     
==========================================
  Hits          280      280              
- Misses        565      626      +61     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[jo-mueller]

Hi @psobolewskiPhD , I just wanted to test this, but for lack of being able to upload zarr images to napari in the first place (ome/ZarrReader#95), it's a bit tricky for me.

I tried to see what would happen with "regular" image data and find that I get an error spamed (see below), with the plugin loading the image fine, after all. That's the traceback:

---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
File [c:\Users\Johan\mambaforge\envs\napari-omero\lib\site-packages\napari\_qt\threads\status_checker.py:114](file:///C:/Users/Johan/mambaforge/envs/napari-omero/lib/site-packages/napari/_qt/threads/status_checker.py:114), in StatusChecker.calculate_status(self=<napari._qt.threads.status_checker.StatusChecker object>)
    110     return
    112 try:
    113     # Calculate the status change from cursor's movement
--> 114     res = viewer._calc_status_from_cursor()
        viewer = Viewer(camera=Camera(center=(7.0, 201.5, 325.5), zoom=1.6377300613496932, angles=(0.0, 0.0, 90.00000000000001), perspective=0.0, mouse_pan=True, mouse_zoom=True), cursor=Cursor(position=(0.0, 6.999999999999995, 290.9530861657729, 444.56724217099224), scaled=True, style=<CursorStyle.STANDARD: 'standard'>, size=1.0), dims=Dims(ndim=4, ndisplay=3, order=(0, 1, 2, 3), axis_labels=('0', '1', '2', '3'), rollable=(True, True, True, True), range=(RangeTuple(start=0.0, stop=0.0, step=1.0), RangeTuple(start=0.0, stop=14.0, step=1.0), RangeTuple(start=0.0, stop=403.0, step=1.0), RangeTuple(start=0.0, stop=651.0, step=1.0)), margin_left=(0.0, 0.0, 0.0, 0.0), margin_right=(0.0, 0.0, 0.0, 0.0), point=(0.0, 7.0, 201.0, 325.0), last_used=0), grid=GridCanvas(stride=1, shape=(-1, -1), enabled=False), layers=[<Image layer 'TL Brightfield' at 0x188f571b9d0>], help='use <2> for transform', status='Ready', tooltip=Tooltip(visible=False, text=''), theme='dark', title='napari', mouse_over_canvas=True, mouse_move_callbacks=[], mouse_drag_callbacks=[], mouse_double_click_callbacks=[<function double_click_to_zoom at 0x00000188E0F9AF80>], mouse_wheel_callbacks=[<function dims_scroll at 0x00000188E0F9A5F0>], _persisted_mouse_event={}, _mouse_drag_gen={}, _mouse_wheel_gen={}, _keymap={})
    115 except Exception as e:  # pragma: no cover # noqa: BLE001
    116     # Our codebase is not threadsafe. It is possible that an
    117     # ViewerModel or Layer state is changed while we are trying to
   (...)
    120     # from crashing the thread. The exception is logged
    121     # and a notification is sent.
    122     notification_manager.dispatch(Notification.from_exception(e))

File [c:\Users\Johan\mambaforge\envs\napari-omero\lib\site-packages\napari\components\viewer_model.py:573](file:///C:/Users/Johan/mambaforge/envs/napari-omero/lib/site-packages/napari/components/viewer_model.py:573), in ViewerModel._calc_status_from_cursor(self=Viewer(camera=Camera(center=(7.0, 201.5, 325.5),...use_drag_gen={}, _mouse_wheel_gen={}, _keymap={}))
    571 active = self.layers.selection.active
    572 if active is not None:
--> 573     status = active.get_status(
        active = <Image layer 'TL Brightfield' at 0x188f571b9d0>
        self = Viewer(camera=Camera(center=(7.0, 201.5, 325.5), zoom=1.6377300613496932, angles=(0.0, 0.0, 90.00000000000001), perspective=0.0, mouse_pan=True, mouse_zoom=True), cursor=Cursor(position=(0.0, 6.999999999999995, 290.9530861657729, 444.56724217099224), scaled=True, style=<CursorStyle.STANDARD: 'standard'>, size=1.0), dims=Dims(ndim=4, ndisplay=3, order=(0, 1, 2, 3), axis_labels=('0', '1', '2', '3'), rollable=(True, True, True, True), range=(RangeTuple(start=0.0, stop=0.0, step=1.0), RangeTuple(start=0.0, stop=14.0, step=1.0), RangeTuple(start=0.0, stop=403.0, step=1.0), RangeTuple(start=0.0, stop=651.0, step=1.0)), margin_left=(0.0, 0.0, 0.0, 0.0), margin_right=(0.0, 0.0, 0.0, 0.0), point=(0.0, 7.0, 201.0, 325.0), last_used=0), grid=GridCanvas(stride=1, shape=(-1, -1), enabled=False), layers=[<Image layer 'TL Brightfield' at 0x188f571b9d0>], help='use <2> for transform', status='Ready', tooltip=Tooltip(visible=False, text=''), theme='dark', title='napari', mouse_over_canvas=True, mouse_move_callbacks=[], mouse_drag_callbacks=[], mouse_double_click_callbacks=[<function double_click_to_zoom at 0x00000188E0F9AF80>], mouse_wheel_callbacks=[<function dims_scroll at 0x00000188E0F9A5F0>], _persisted_mouse_event={}, _mouse_drag_gen={}, _mouse_wheel_gen={}, _keymap={})
        self.cursor.position = (0.0, 6.999999999999995, 290.9530861657729, 444.56724217099224)
        self.cursor = Cursor(position=(0.0, 6.999999999999995, 290.9530861657729, 444.56724217099224), scaled=True, style=<CursorStyle.STANDARD: 'standard'>, size=1.0)
        self.cursor._view_direction = array([0.00000e+00, 1.00000e+00, 1.46562e-15, 1.94476e-14])
...
    656     return self._calculate_value_from_ray(values)
    658 return None

IndexError: too many indices for array: array is 2-dimensional, but 3 were indexed

It seems like it has something to do with how some of the viewer/camera parameters are set and the dimensionality of the data itself?

[psobolewskiPhD]

Hi! You don't need a zarr in OMERO just a pyramidal image. We have a lot of slide scans in ours.
If you can upload to your OMERO instance, you could grab an image from e.g. here:
https://qupath.readthedocs.io/en/stable/docs/intro/acknowledgements.html

I think the error you are getting looks like:
#66
Which is upstream issue: napari/napari#7407
Which I fixed 🎉 so should be resolved in 0.5.6

Meanwhile the fix is to not mouse the canvas until the layer is fully loaded.

Importantly, if it's not a pyramidal image, none of the code in this PR should be triggered:
it's gated by if image.requiresPixelsPyramid():