for MCF10A also seems to be dependent a lot on density! looks very different compared to this (over)confluent epithelia i imaged some time ago on confocal. +1 for zarr+dask, has been working great for us with on data 👌 any tricks on how to convert .czi to .zarr without loosing compression?

In my experience the default zarr already scores quite efficiency on disk, but thats for data that has lots of zeros. In terms of czi conversion i would use the bioio package. However i don't know if you can maintain the compression. Otherwise native zarr supports zipstore, which might help.

Nice! Indeed nuclei are pancakes when they are not inside a 3D matrix, since glass or hard plastic is incredibly charged, pulling all towards the interface. Regarding the big files, i suggest to use zarr and dask, seems like the whole field is moving there.

Thanks to Hanjin Liu, napari now has a command palette! 🎨 This thing is really nice to use — the core team can no longer live without it — and it even works with plugin commands! 🤩