diff --git a/README.md b/README.md
index eb11c4f..3e0eec9 100644
--- a/README.md
+++ b/README.md
@@ -25,7 +25,14 @@ object with the following properties:
 -   `lsid` must be set to the location of the Zarr data:
     - the absolute path to the dataset if available via the filesystem
     - an URI defined as `s3://s3.<region>.amazonaws.com/<bucket>/<path/to/dataset.zarr>`
-      for datasets stored on Amazon  S3
+      for datasets stored on Amazon S3
+
+The Zarr dataset must be conformant with the default layout generated by `bioformats2raw` i.e.:
+
+-    the version of the NGFF specification must be 0.4
+-    the image and the Zarr multiscales group must have 5 declared dimensions
+-    the chunks must be stored in the (t, c, z, y, x) order
+-    the resolution level must be downsampled along the (y, x) axis
 
 Permissions for accessing Zarr dataset hosted on S3 buckets can be managed: