Using latest ngff-zarr:
$ pip freeze | grep zarr
ngff-zarr==0.45.0
zarr-metadata==0.5.0
zarrista==0.1.0
With this script to read the image at https://ome.github.io/ome-ngff-validator/?source=https://livingobjects.ebi.ac.uk/idr/zarr/v0.5/idr0062A/6001240_labels.zarr
where the largest resolution has:
Shape: 2,236,275,271
Chunk: 1,1,256,256
Shard: 1,10,512,512
import ngff_zarr as nz
multiscales = nz.from_ngff_zarr('https://livingobjects.ebi.ac.uk/idr/zarr/v0.5/idr0062A/6001240_labels.zarr')
nz.to_ngff_zarr("6001240_shards.ome.zarr", multiscales, chunks_per_shard={"z": 10, "y": 2, "x": 2, "c": 2})
The resulting image has treated the x and y chunk sizes quite differently.
Instead of the chunks being "square", I see
Chunk: 1,1,55,271
Shard: 2,10,275,271
Maybe this is the expected behaviour (because the image is kinda small?), but I just wanted to check / understand a bit better?
Thanks!
Using latest ngff-zarr:
With this script to read the image at https://ome.github.io/ome-ngff-validator/?source=https://livingobjects.ebi.ac.uk/idr/zarr/v0.5/idr0062A/6001240_labels.zarr
where the largest resolution has:
The resulting image has treated the
xandychunk sizes quite differently.Instead of the chunks being "square", I see
Maybe this is the expected behaviour (because the image is kinda small?), but I just wanted to check / understand a bit better?
Thanks!