Peter Sobolewski

I can reproduce this. Using the script in the OP painting in the tensorstore array is slow, but not as bad as in the video. But when I use the...

I tested with a zarr on disk without tensorstore, using: `z2 = zarr.open('test_zarr.zarr', mode='w', shape=(2, 2000, 2000), chunks=(1, 1000, 1000), dtype='i4')` Performance of painting felt similar to tensorstore and the...

I was playing with this pattern: https://google.github.io/tensorstore/python/api/tensorstore.Transaction.html Script: ``` import numpy as np import napari import tensorstore as ts import zarr array_size = (2,2000, 2000) np_array = np.zeros(array_size, dtype='uint32') zarr_path...

Playing a bit further, not sure if this is too hacky, but it does work: paint, commit, refresh transaction, point layer data at the new transaction (just refreshing doesn't suffice):...

I assume you could go on painting indefinitely before doing the commit, so it's actually not bad from a UX point of view. It's just `save` basically, consisting of commit,...

@DevinBayly curious if you got this (or egl one) to work using apptainer? (with nvidia GPU acceleration)

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...

Echoing Will's comment, can you clarify if you are using napari >0.5 and have async on? See my OP for comparisons with and without. Either way need to make sure...

@jo-mueller I think that's more of a napari issue, see for example: - flashing when zooming in and out: https://github.com/napari/napari/issues/7391 - smart pre-fetch: https://github.com/napari/napari/issues/2350 Edit: I pushed an update to...

@jo-mueller I've used this branch quite it a bit and seems good. I think we could merge, release and iterate? In terms of a release, I think dealing with the...