Field Rotation Attribute

[ajpatterson]

Moved from ​http://www.ome-xml.org/ticket/116

Request for "Field Rotation" attribute.
See Forums, Data Model > "OME schema parameters" topic

"Field rotation - the field of view/capture can be rotated to frame an image better or for consistency (e.g. always imaging a zebrafish with head to the left and tail to the right of the image with spine horizontal). Not particularly important to record for single images but useful for stitched images and tracking.

Field rotation is measured in degrees. The Leica LAS AF software (.lif files) include this, they refer to it as the scan field rotator and scan rotation is measured in degrees (-100 to +100)."

LIF file XML: (Rosier-Image007.xml)
Wonder if the rotation is always positive number, and rotation direction is either +/- 1? Need more examples.

<HardwareSetting>
  <FilterSetting>
    <FilterSettingRecord ObjectName="Scan Field Rotator MA4" ClassName="CRotator" 
        Attribute="Scan Rotation" Description="Scan Field Rotation" Data="0" Variant="77"
        VariantType="5" Unit="degrees">
    </FilterSettingRecord>
    <FilterSettingRecord ObjectName="Scan Field Rotator MA4" ClassName="CRotator" 
        Attribute="Rotation Direction" Description="Rotation Direction" Data="0" Variant="1" 
        VariantType="3">
    </FilterSettingRecord>

[ajpatterson]

Editing description just to allow for better printing in report.

[jburel]

moving to 4.4 for consideration

[ajpatterson]

Decision taken in meeting today with Jean-Marie & Roger to implement this as a general affine transform on image rather than just a rotation angle.

[wmoore]

My only concern with this idea is that it's not at all clear to a user that they should store a rotation angle of N degrees (which is a scanner / scan setting) by calculating an affine transform. Especially since the affine transform may be used for a bunch of other stuff. Seems this is mixing acquisition metadata with visualisation / post-processing settings in a way that is not easily 'unmixed'.
I guess there's probably no harm in adding an affine transform, so that when viewed, the image appears as the user expects (I wonder if that's what the Leica software does??).
But do we have any plans to support affine transforms when viewing images in our clients? This would have to done server-side, right?
And I think you'd still want to know the acquisition settings too.

I wonder if it's still a requested feature for that user (from 4 years ago)? http://www.openmicroscopy.org/community/viewtopic.php?f=15&t=342

[ajpatterson]

Roger raised the idea of allowing the storing of either a rotation, a scaling, or a matrix.

We decided that expanding the transform complex type and putting in the appropriate helper/access functions would be a separate problem.

[wmoore]

The 'rotation, scaling or matrix' are still visualisation / processing attributes. E.g. this is an image that you've "stitched" together with other images and you want to record how to transform it into the stitched coordinates.
But this is different from recording the acquisition parameters.
E.g. If I'm looking at the acquisition parameters in the right-panel of the clients, so that I can set the microscope up in the same way again etc. This won't show up there.

[ajpatterson]

Changes in ​https://github.com/openmicroscopy/bioformats/pull/424

[mlinkert]

Change reverted in ​https://github.com/openmicroscopy/bioformats/pull/467

Roger's documentation comment from that PR should be addressed when this work is completed.

[jamoore]

Referencing ticket #10284 has changed sprint.

[jamoore]

Referencing ticket #10284 has changed sprint.