encode an array of these values rather than the row/column size
encode the paths to the individual images
is probably the easiest for the moment. 3. is more verbose but is probably my long-term preference.
@sbesson
"The contents of the Field_* folder should probably be multiscales. That adds yet another layer to this already pretty deep hierarchy, but it will mean we can always assume a pyramid."
Also assuming multiscales gets increasingly used for thumbnailing, big 👍 for having this a MUST rather than a SHOULD in the spec from my side.
Comments from now closed PR #28
@joshmoore Feedback from https://github.com/ome/ome-zarr-py/pull/56 : The contents of the Field_* folder should probably be multiscales. That adds yet another layer to this already pretty deep hierarchy, but it will mean we can always assume a pyramid. Several of the string elements in the path aren't readably computable at the moment (see https://github.com/ome/ome-zarr-py/pull/56/files#diff-b50d9715cc6e4017cfc055fd0ed73ecb5d9158e17f4d58ca5b3ba08b89c46657R354-R358) Ideas on this front are:
@sbesson "The contents of the Field_* folder should probably be multiscales. That adds yet another layer to this already pretty deep hierarchy, but it will mean we can always assume a pyramid." Also assuming multiscales gets increasingly used for thumbnailing, big 👍 for having this a MUST rather than a SHOULD in the spec from my side.