My experiments so far seem to indicate that I can retrieve images covering the whole PFS FOV with a single request. I'll continue monitoring this; if problems turn up in the future, breaking down the request into tiles of 1x1 degrees should not be too difficult.

> breaking down the request into tiles of 1x1 degrees should not be too difficult.

That statement may actually be too optimistic: since `retrieve_image` retrieves a region whose borders are not aligned with meridians (at least that's my impression, otherwise images near the poles would be quite distorted), there is no simple way to break down a large patch into smaller ones to be queried individually. So let's hope that SDSS will continue to deliver sufficiently large patches.

On branch "minimal_version" of https://github.com/Subaru-PFS/ets_fiberalloc I have put a very simple demo of the image retrieval in the file skyimage.py.

Retrieval from SDSS and DSS2 seems to work, but PANSTARRS and PS1 appear broken. If we need them as fallback I'll try to do further debugging and am grateful for any hints!

Hi Martin,
 
I would not worry about PANSTARRS too much right now. 
 
If there is also a way to externally supply an image via command line parameter, I would think this covers most usecases.
 
Max

Do we still expect to use this functionality, mxhf? Given how the expected functionality of shuffle in PFS has changed over the last years (it basically just selects suitable guide star candidates and does not re-point the tetescope on its own), we may not need it any more. It's not part of the software that's currently in use, in any case.