RFC5: coordinate systems and transformations #17
Conversation
…ller/ngff-spec into upgrade-to-jupyter-book-2
jo-mueller
added
the
RFC
Automated Review URLs |
affine/rotation matrices should always be 2D
There was a problem hiding this comment.
Trying to use this branch from the validator...
The image.schema has
"$id": "https://ngff.openmicroscopy.org/0.6.dev1/schemas/image.schema",
and
"$ref": "https://ngff.openmicroscopy.org/0.6.dev1/schemas/_version.schema"
https://deploy-preview-48--ome-ngff-validator.netlify.app/?source=http://localhost:8000/4995115_tiles.zarr/4995115_cropped_400_100_400_400_rot10.zarr/
Is currently failing with
can't resolve reference coordinate_transformations.schema from id https://ngff.openmicroscopy.org/0.6.dev1/schemas/image.schema
So hopefully that will be fixed by addressing those lines 🤞
|
@will-moore can you take a look at the schemas over at this PR? I haven't done much work on the schemas here yet (seemed like a big beast to tackle while the spec text was still changing), but now I think I'm almost there. Edit: Oh and I am seeing that I changed all the versionings at said PR to |
|
I think Fixed it to use that branch in ome/ome-ngff-validator@b4ee3bb and this is looking good now, thanks. See |
I'm merging this so that all necessary changes regarding the addition of rfc5-stuff to the main branch of the ngff-spec repo can be in one place. This branch was supposed to be a break-out to keep the commit-history of ome#17 clean. * Update input/output to input_axes/output_axes in schema * Add description and required field to byDimension * Update mapAxis schema to use integer array * Add path and interpolation to displacements schema The displacements object now includes a required 'path' property for specifying the zarr array location and an 'interpolation' property with supported methods. This enhances the schema's ability to describe displacement fields and their application. * Add path and interpolation to coordinates schema Introduces 'path' and 'interpolation' properties to the 'coordinates' object in the schema, specifying the location of the coordinate field and the interpolation method to use. The 'path' property is now required. * only allow paths, no URLs * move required fields to correct places * name musnt't be empty * added axis types to schema * Added descriptions to schemas * updated versions to 0.6dev2 * fix paths and versions * pull schema from correct location * Added action to run the tests * update config reference * update versions in test suite * Revert "update config reference" This reverts commit ff3fedd. * update all version references to "0.6dev2" * update version reference * at least two spatial axes * allow any axis type * update version reference * Add maxItems constraint to axes schema Set a maximum of 5 items for the axes array in the schema to enforce limits on the number of axes allowed. * Refactor image schema for coordinate transformations Refactors the definition of coordinateTransformations and coordinateSystems to use inline array schemas with stricter constraints. Adds a new multiscale_coordinateTransformations definition to support scale and translate transformations for multiscale datasets. * Update to NGFF 0.6dev2 and refactor coordinate systems Updated example and test JSON files to use the NGFF 0.6dev2 specification. Replaced 'axes' with 'coordinateSystems', added explicit 'input' and 'output' fields to coordinateTransformations, and restructured transformation chains for clarity and compliance with the new spec. * fix schema resolution
Co-authored-by: Will Moore <[email protected]> Co-authored-by: David Stansby <[email protected]> Co-authored-by: Davis Bennett <[email protected]>
That's what I had in mind. Thanks! We can also do that at the Hackathon?
Sorry, I probably broke that. I wanted to have everything in one place here to avoid polluting the commit history any worse than it already is :D Changing the schemas for the validator to this place should fix it, though! |
|
@jo-mueller I pushed a fix about the same time as I made that comment (when I realised it was broken). |
2 participants
Built on top top of #14 and #20 (requires these two to be merged first). Contains the changes proposed in the latest versions of rfc5.