[feature wish] Expose scaling parameter in derive_image api

[clacrow]

Would be nice to have a control over image scaling parameters when using derive_image.
Possible use-case is when i’m doing projection i need to set some axis equal to 1, but if there is a scaling with image levels at that axis it will fail.

Related issue in fractal-tasks-core:
fractal-analytics-platform/fractal-tasks-core#991

[lorenzocerrone]

Hi @clacrow,

In general, I would prefer to narrow rather than expand the derive_image API.

Exposing more parameters in the image derivation, such as scale, could lead to unforeseen issues since most tasks (besides projections) rely on the fact that derived images have the same shape at all levels.

In those cases, I would encourage you to use the create_empty_ome_zarr function instead of the derive.
I can see how scale would be useful in certain contexts, so I keep this open to see if more people would be interested in this feature.

[lorenzocerrone]

Hi @clacrow,

Ngio 0.5 now supports creating images with inhomogeneous scaling. There is no need now to change the scaling when squeezing an axis dimension to 1.

In the end, we decided not to expose the scaling to the derive API, so for now, if you want to explicitly change the scaling factor in an image, the only option is to use the creation APIs.