RIQ FCa : Derotator usage for IFU image

[astronomyk]

https://jira.eso.org/browse/MET-2136

When using the derotator to turn the image field by 90 degrees and then reconstruct an IFU image: what is the accuracy of the derotator motion ? If non-negligible, does the pipeline take into account this in the error propagation ?
Later we can read: "One of the major challenges is to find the exact centre of rotation" (Sect. 8.9, p.p153)

@sesquideus what was the conclusion of the chat with Roy about the accuracy of the derpotato?

[sesquideus]

@astronomyk That we should not assume that the difference in angles is 90° exactly, or below our detection limit. While I think the derotator hardware team promises the precision repeatability is on the order of arcseconds, Roy thought that getting rid of that assumption would be nice. This was also my primary motivation for designing the algorithm and Roy liked that.

Nevertheless this is out of the critical path, the data-to-model-pixel-overlap algorithm is already working. It can be even very easily extended for non-rectangular pixels – Roland Ottensamer told me that when they wrote the Herschel pipeline they also counted with distortions in the end, even if it was not originally planned. Their "data pixels" are general quadrilaterals that are later resampled onto a square grid.

We only need to know the exact position and orientation of each exposure. I don't suppose we are able to determine it ourselves, but if it is in the input data, the algorithm can process it.

[astronomyk]

@sesquideus thanks for the summary for the discussion with Roy. Can you come up with an answer for Faustine please?

Feel free to answer her directly on Jira too