Processing JWST images

[Mike Cuffe]

I'm processing the publicly available raw images from the James Webb Space Telescope. The raw mono images are taken using different wavelength filters and by assigning different colours to each filter a full colour image produced. The problem I have is that some filters produce images that are half the size of others, so in order to overlay them they have to be stretched then aligned using 'Align Layers by Stars' but I am struggling to get the registration perfect. Is there a better way to stretch and align these images?

[James Ritson]

On 12/14/2022 at 3:45 PM, Mike Cuffe said:

I'm processing the publicly available raw images from the James Webb Space Telescope. The raw mono images are taken using different wavelength filters and by assigning different colours to each filter a full colour image produced. The problem I have is that some filters produce images that are half the size of others, so in order to overlay them they have to be stretched then aligned using 'Align Layers by Stars' but I am struggling to get the registration perfect. Is there a better way to stretch and align these images?

Hey Mike, sounds like some of the data is 2x2 pixel binned? You should actually be able to select both 1x1 and 2x2 layers (pasted into one document but maintaining their respective resolutions) and use the star alignment method—it's supposed to scale and translate layers where required.

Do you have a working example where this is failing? Is star detail rendered differently, e.g. are they bloated in some of the images but not others?

If you have to use a manual process, I would probably change one of the layer blend modes to Difference, then scale and rotate until the blending between the layers is almost black (i.e. "no difference"). Another approach is to quickly drop the opacity of one layer down (you can use number keys, e.g. 5 for 50%, 0 for 100%) so you can see both source and destination layers.

Hope that helps!

[Mike Cuffe]

Hi James, thanks for your reply.

As I'm sure you know the zip files downloaded from the Mast Portal are very large so not practical to attach here. So here's the contents of the Manifest HTML docs which came with the fits files I downloaded for Stephan's Quintet, I hope this gives you enough data to download the data yourself.

mast:JWST/product/jw02732-o001_t001_nircam_clear-f090w_i2d.fits    JWST/jw02732-o001_t001_nircam_clear-f090w/jw02732-o001_t001_nircam_clear-f090w_i2d.fits    PUBLIC    OK    anonymous

mast:JWST/product/jw02732-o001_t001_nircam_clear-f200w_i2d.fits    JWST/jw02732-o001_t001_nircam_clear-f200w/jw02732-o001_t001_nircam_clear-f200w_i2d.fits    PUBLIC    OK    anonymous

mast:JWST/product/jw02732-o001_t001_nircam_clear-f444w_i2d.fits    JWST/jw02732-o001_t001_nircam_clear-f444w/jw02732-o001_t001_nircam_clear-f444w_i2d.fits    PUBLIC    OK    anonymous

mast:JWST/product/jw02732-o001_t001_nircam_clear-f356w_i2d.fits    JWST/jw02732-o001_t001_nircam_clear-f356w/jw02732-o001_t001_nircam_clear-f356w_i2d.fits    PUBLIC    OK    anonymous

mast:JWST/product/jw02732-o001_t001_nircam_clear-f150w_i2d.fits    JWST/jw02732-o001_t001_nircam_clear-f150w/jw02732-o001_t001_nircam_clear-f150w_i2d.fits    PUBLIC    OK    anonymous

mast:JWST/product/jw02732-o001_t001_nircam_clear-f277w_i2d.fits    JWST/jw02732-o001_t001_nircam_clear-f277w/jw02732-o001_t001_nircam_clear-f277w_i2d.fits    PUBLIC    OK    anonymous

This set illustrates the problem I've found on all JWST data I've so far downloaded, filters f090w, f150w and f200w are approximately 14000 x 12000 whereas filters f277w, f356w and f444w are approximately 7000 x 6000.

I started off trying 'Align layers by stars' in the way you describe, I'd have to try it again, but I think I found that it would scale and align the bigger images to the smaller but it seems to struggle the other way round i.e. to align the smaller images to the bigger ones.

So far my most successful method is to scale up the smaller files x2 to roughly the same size as the bigger ones, do a 'by eye' alignment (blend 50%) then let 'Align layers by stars' do the final alignment, it works (mostly) but it would be so much easier if all the data was the same size. If you think I should be able to download all filters in the same resolution then I'd love to know how to do it.

Thanks for your help with this.

Mike