Investigate (x,y) Ordering Between RSIM and ImVIZ

[PaulHuwe]

RSIM and ImVIZ are transpositions of each other (x <-> y), resulting in images like this:

Which should have a shape more like:

We need to investigate if RSIM is using the wrong convention, or if this needs to incorporated on the Imviz side.

[bmorris3]

Thanks for the heads up. Generating some catalogs and testing on my end too.

[bmorris3]

Still looking into this. I can get sensible SCA orientations if I do a particular set of transformations to each SCA's data, similar to an x orientation flip, y orientation flip, and a transpose. I can't figure out if these are sensible, but the cluster looks better:

[schlafly]

To try to make progress here... can we pick a particular star on these images and make sure that imviz is putting the star where it should be?

This file was generated from a catalog. We should be able to plot the positions of the stars in that catalog on top of this image and see if they match. e.g., something like:

im2 = asdf.open('l2-cluster-smaller-sca7.asdf')['roman']['data'].value
cat = Table.read('cat-270-66-cluster-smaller.fits')
wcs = asdf.open('l2-cluster-smaller-sca7.asdf')['roman']['meta']['wcs']
xx, yy = wcs.world_to_pixel(cat['ra'], cat['dec'])
m = (np.abs(yy - 2048) < 2048) & (np.abs(xx - 2048) < 2048) ; plot(xx[m], yy[m], 'r+')

That nicely produces points on the sources in the image. That tells me that the simulator WCS & images are consistent, so I think this has to be on the imviz side?

It's easy for me to imagine (x, y) <-> (y, x) issues; there's the usual one with (x, y) corresponding to image[y, x]. I claim that I'm following the accepted python / imshow / wcs API conventions there, but if you think otherwise, let me know. I don't have a picture for how we could make an error like im -> im[::-1, ::-1]; that's much harder.

If it's useful to have these files, I've dropped them on /grp/roman/eschlafly/romanisim/20230420.