jluethi
commented
1 year ago Actually, this should be a fairly rare edge-case for the moment: We mostly want to know edge of FOV information and most FOVs we currently have actually have a size that is divisible by 2 many times.
It will come up more though potentially when/if we start doing registration per FOV and change the FOV size to the consensus size
When label images need to get upscaled for measurements in the napari workflow wrapper and they are not a simple 2x (4x/8x/etc) downscaled (e.g. in the search-first case, see https://github.com/fractal-analytics-platform/fractal-tasks-core/issues/201), the upscaled label images get padded with 0s. In that case, some edge detection measurements on the label images may be off.
This is a backlog issue to document this edge case.