Do centering of HCI exposures on full cube?
From communication with Pierre Baudoz;
Note that to limit the effect of noise, the recentering with the waffle might need:
- to take the estimated centroids of the whole cube
- to do a fit of the positions of each psf assuming the drift does not contain high temporal effect (if there is vibrations, it might not be the case and we should use each estimations)
- then recenter each images using the fitted position of the coronagraph Same for photometric estimation using the waffle (especially since pyramid might introduce photometric variations)
The current HCI pipeline is entirely per exposure. I'm inclined to mark this as beyond the scope of our pipeline.
@verdoes what do you think? Please assign to me if you think we should do this now, otherwise make it unassigned so we can decide to do it later if we wish.