StarTools

Fully post-processed in ST 1.9a. All 6 filters used, basically mapped to the same false color hues the same way NASA did. Their version (unsure how they did it, what they used, or how selective they may have been. Though that's perhaps unavoidable trying to mash 6 filters together, and I'm sure my attempt is artsy as well) can be seen here: https://stsci-opo.org/STScI-01GFXR207QM ... QSF3CT.png

This was rather difficult, but I finally got something out of the data. For sure a work in progress, as I made a number of mistakes. What's the SS-Airy Disk % for JWST anyway?

For now, in short, I composited three of the files as RGB (though I used a NB matrix), then used Layer module operations to bring in the other three files, which I had already processed through to matching cropped/binned final images, "mono" but colorized in the Color module to the "correct" hue.

As you can imagine, JWST doesn't quite fit the filter choices in our new composite module options, though it seems your choice there will send you on the path to having green cap or purple cap. Though with funny data like this, I kind of wanted to be able to pick either or both once I got there.

I'll keep trying some different versions of this (maybe changing up my luminance weighting in compose a bit more too), but I'd like to see what others can make of it too.

JWST data can be downloaded from MAST Archives, but usage is a bit awkward. The fits files have multiple images contained within, so you have to extract out the appropriate panel/layer. I've used ASTAP for that before, but this data crashed it. Fits Liberator is ok, but had trouble with the 90 filter, and only saves out as tiff. So I ended up using Gimp.

Then, some JWST filters are taken bin 1, others bin 2, so you sort of have to scale some down to match.

And then there's star registration. No, NASA did not do that for us. This time I couldn't get anything to register/star align these files. ASTAP failed. And, after I cropped edges so each file had the same xy, DSS failed also. So, as I clearly hate myself, I picked a reference file and aligned each file by eyeball/hand in Gimp - using levels stretched images to figure out the scaling, rotation, and xy displacement, and then applying that to a linear version of the file and saving it back out.

Maybe someone can try PI perhaps and see if that program can star align the files?

My extracted, scaled (ballpark 50% for the bin1 files), cropped, and hand-registered files are here in the so-named folder: https://drive.google.com/drive/folders/ ... sp=sharing

There's also a folder linked that's just extracted, scaled, and cropped, but not yet registered, again if anyone can manage to get a program to align them. I'm sure my eyeballed ones are flawed since I could only be so "perfect," and I couldn't do interpolation either.

And again, the originals could be obtained from MAST if you want to start from scratch.

I think I got them to align using APP. Not sure as I am still learning the program. Link to files https://www.dropbox.com/scl/fo/0mz67j6v ... drhleayaas
Che

Cheman wrote: ↑Mon Dec 12, 2022 10:57 pm I think I got them to align using APP. Not sure as I am still learning the program. Link to files https://www.dropbox.com/scl/fo/0mz67j6v ... drhleayaas
Che

Thank you for the effort, Che!

I forgot about APP, and had figured this was probably a job for PI's pre-processing. Though I admit I didn't try Siril either. I use Siril for all sorts of handy things, often for quick looks at data, but just plain stacking has always befuddled me - let alone registering finished stacks. So I have no clue how to even go about it.

That said - I downloaded your share and tried the files. They seem to be identical to the ones I registered by hand. I know I'm not that good!

For checking I just used RGB composite in Siril to load pairs as R and G, then blinked them (except for filter 90, in case that was the reference file, which it was for me). Even zoomed way in I could see no difference blinking between the channels. I then looked at the color composite, and could find no differentiation of pixels. I tried that for each file. Then as a sanity check I loaded one of my hand registered files to the unregistered version (as uploaded to that folder), and for that the blink showed a clear difference in scale, rotation, and displacement, and the color composite also revealed the difference.

So, not sure what happened and dunno if we can congratulate APP just yet. Hope so though, that would be cool.

Mike in Rancho wrote: ↑Mon Dec 12, 2022 7:07 pm Fully post-processed in ST 1.9a.

Oh wow! Very nice! That's pretty darn close color-wise to the NASA image.

I wonder how/why they got so much Paschen-alpha signal (mapped to blue), even though you endeavored to accomplish the same?

admin wrote: ↑Tue Dec 13, 2022 7:40 am Oh wow! Very nice! That's pretty darn close color-wise to the NASA image.

I wonder how/why they got so much Paschen-alpha signal (mapped to blue), even though you endeavored to accomplish the same?

Ah yes, Paschen-alpha, of course.

Well Ivo you sent me on an internet hunt. So, it seems this is akin to the Balmer series (Ha, Hb...) we are all used to in visual, except now in IR. Paschen alpha being the photon released with a 4 to 3 electron state transition, coming in at 1875 nm. Which would ever so nicely match up with JWST's NIRCAM 187N (microns, narrow) filter. Gee, seems maybe they chose these filter bands on purpose. Who knew? And it's our old friend hydrogen to boot!

And yep that's mapped to blue, and so I chose to include that for my initial three-channel RGB base image - though I put 335M yellow into G, and 444W orange into red (which being wide it covers anyway) so that I could use the 470 for really red highlights later.

There's actually quite a bit of such blue strewn about, though perhaps I lost some of it in my chosen optidev. I think some pre-thought about luminance exposure settings for compose is probably in order. Likely I had the other two channels relatively too bright.

Then the next matter is how to get to 6 filters. And, if layering, what layer operation to choose, how strong, and in what order.

The 90W filter also has substantial data, but at the claimed map-to-purple, I see little of that in NASA's image. When I brought that in, purple dominated and/or altered the nice blue I had from the 187N. So in the end I just destretched the 90W filter to suppress the purple takeover.

Layering the 200W (cyan) in afterwards can also tame down that purple, but this filter picks up so many stars that it needs to be suppressed also, and so that effect is minimized.

470N as red seemed the easiest to try to almost replicate what NASA did. Brought in last as a lighten layer, it really powers up that handful of bright glowing red regions, and adds surface detail to otherwise very dark shadow nebula regions.

But where did NASA hide 90W? It's really quite striking. Perhaps overstretched a bit here, as there could be a bit of artifacting bands in the upper left when oriented like this. But even so, if they used it, we should see more purple.

And capping (I tried) after the blend doesn't turn the purple into a nice blue, but more of a pukey grey.

Nonetheless, the sharpness of the focus and diffraction spiking at this image scale is amazing.


I'll try some different combos. Fun to play with. And for the base image I'll also try to get SS right next time if I use it. I think I should have moved airy disk all the way right instead of all the way left.

Technically a bit over my head i guess, but anyhow the first image looks great Mike. Great job !

Mike in Rancho wrote: ↑Tue Dec 13, 2022 6:09 am

Cheman wrote: ↑Mon Dec 12, 2022 10:57 pm I think I got them to align using APP. Not sure as I am still learning the program. Link to files https://www.dropbox.com/scl/fo/0mz67j6v ... drhleayaas
Che

So, not sure what happened and dunno if we can congratulate APP just yet. Hope so though, that would be cool.

I'm thinking it was my lack of knowledge of the program rather than the program itself as its very new to me. The first time it failed and the next I tried your self aligned frames so it probably failed as well and just gave me back what I put in. I use siril for most of my stacking, but I use modified scripts to perform the action. I dont know if it can handle these files or not. I do love that program tho. You could post on their forum and APP for that matter and would probably get the devs to respond

Mike in Rancho wrote: ↑Tue Dec 13, 2022 8:49 am So, it seems this is akin to the Balmer series (Ha, Hb...) we are all used to in visual, except now in IR.

Yup. Astronomers being astronomers, things are - once again - wonderfully inconsistent. It'd be so much easier if would just call H-alpha "Balmer-alpha". It's all hydrogen, so the "H" is IMHO fairly useless/misleading in the context of the other series. Plenty of other examples ("H-II", "Planetary nebula", etc.) where it almost seems as if the aim was to obfuscate or confuse.

I'll try some different combos. Fun to play with. And for the base image I'll also try to get SS right next time if I use it. I think I should have moved airy disk all the way right instead of all the way left.

It'd be really cool if the good folks at NASA would publish their workflow for processing, so the resulting images can be scrutinized and validated. The "pretty picture" outreach efforts should not suffer from that at all.

Regardless, it's amazing us mere mortals can explore infrared spectrum imagery so easily these days.

Mike
I tried a few more settings in APP using the original files. It always fails reg on the f090w file. the others all say registered. the number of stars detected on that one is 254 while the others are all over 1500. I am assuming that this is the problem but have not been able to find a setting that will successfully register all. I might try a few more settings as it's a learning experience but am runnin out of ideas. If you would like, I can upload all the others
Che.

Cheman wrote: ↑Wed Dec 14, 2022 3:26 pm Mike
I tried a few more settings in APP using the original files. It always fails reg on the f090w file. the others all say registered. the number of stars detected on that one is 254 while the others are all over 1500. I am assuming that this is the problem but have not been able to find a setting that will successfully register all. I might try a few more settings as it's a learning experience but am runnin out of ideas. If you would like, I can upload all the others
Che.

I changed the reference frame and made a few other settings adjustments and was finaly able to get APP to run without errors. I also included a file of all frames integrated together give it a try and see if I was sucessful this time. https://www.dropbox.com/scl/fo/eu5xuhjc ... k5j7vuk0so