Astro to Astro-art Continuum, 2024 Edition

[decay]

Hi Ivo & Martin, really, that’s far too kind . Of course, ST is greatly designed to be used without nuts and bolts, bits and bytes and even without too many graphs, diagrams or whatever. But AP really isn’t a simple subject and I think diving a bit deeper can help to get a better understanding of what we are doing. And some bits and bytes and graphs and diagrams might be useful to populate Mike’s Astro-art Continuum. Hopefully

Martin, I guess getting older also gives the chance to learn and to do new things in life. You decided to go for AP and this is a fantastic great world! And there is absolutely no need to dive too deeply into computers, electronics and maths if you don’t want to. (Which is perfectly fine, of course!)

My youth was in the 80s and I grew up with electronics, soldering and coding the early home computers. The 80s were probably a good decade to learn the basics of all this …

Dietmar.

[Mike in Rancho]

Due to another discussion about NB palette options, I was doing some googling and came across this article that summarizes steps taken by the pros who process JWST data. https://webbtelescope.org/contents/arti ... mages-made

They also mention it's very similar to what is/was done with Hubble data, so we could probably consider it mostly applicable to visual spectrum images (whether broadband or the usual suspect narrowband filters) as well.

I found it fairly interesting what they deem amenable for healing, repair, touch-up, etc., but that is still a properly accurate representation of the science. And also that the responsible researchers are given veto or suggestion power over the processing, to make sure the image backs up their science and discoveries.

I haven't yet found exactly what processing tools they find permissible, though I get the sense all the usual suspects would be allowed.

And while they do certain things with an aim to showcase particular scientific matters (while still being "engaging"), it seems they are layering together different non-linear stretches of the various filter channels - at least for IR and narrowband. Of course JWST is all IR, so I'm a little more interested in how they might have composted HST SHO. Still trying to dig that up for Hubble, but if the workflow here is the same basic assembly steps, I'm thinking that Hubble (say, Pillars of Creation) may not necessarily have maintained relative NB ratios.

[decay]

Interesting article, Mike. Thanks for pointing out. The process description starts quite detailed, explaining how the images are aligned and stretched. And most of the other covered topics are described detailed, too. But I wonder if and what is missing. The process of “bringing out the details” (section “Editing for Composition”) is covered rather vaguely. How do they do that? For example, do they use local contrast enhancements, like we discussed before (Contrast, Sharp, HDR modules)? Do they use deconvolution or something like the Super Structure module does?
All these explanations regarding “Removing Artifacts” and the orientation of the diffraction spikes are interesting, but other (missing) topics may be more important?

If they really use FITS Liberator for stretching, like depicted, stretching seems to be rather conservative, using Asinh function (don’t know what ‘Pow’ means. The mathematical “Power” function?). OptiDev is probably more sophisticated.

I'm thinking that Hubble (say, Pillars of Creation) may not necessarily have maintained relative NB ratios.

I also think that they may not maintain relative NB ratios. I know that Ivo strongly recommends to maintain relative NB ratios and doing not would reduce or even eliminate scientific value and claim. But I wonder if this isn’t just a simple shift in our continuum towards art. And not a quantum leap. Using non-linear global stretching curves compromises the evaluation of brightness levels. Using local contrast enhancements it is not longer possible even to compare brightness levels at different locations of the image. And using different stretching curves for each component – well it additionally compromises relative NB ratios. Probably just one more step towards art?

Best regards, Dietmar.

[Mike in Rancho]

Hi Dietmar, thanks for adding your perspective.

I agree with a number of your takes. I would like to see just how they enhance the images, nuts-and-bolts description. And yeah, are they using the same FITS Liberator that I have? Mine is so limited it almost seems worthless, other than reading FITS headers. Though maybe I just don't know how to use it. But, the article does say that after the files are stretched, they take them all into more professional software. They don't say which, just add that Gimp is a nice free program.

I don't think I would say it's a step towards "art," though that all depends on definition and scope. The JWST team does state that things become more subjective. And that's okay! To me the art side is more manipulative, even misrepresentative, alterations, if serving only aesthetic purpose. Like mask subtractions, Forax, and so on. As to the X-tools...nobody knows!

I did come across one page that includes more HST discussion (it takes forever to load but eventually does): https://asd.gsfc.nasa.gov/blueshift/ind ... lse-color/

Also, this page is supposedly the blog of the professional image processors? https://illuminateduniverse.org/

It's a little weird, and I haven't yet read through any of the links/write-ups to see if there are any disclosures about curve stretching, contrast, HDR, denoise, or even sharpening (and if so, what type!?).

[decay]

I don't think I would say it's a step towards "art," though that all depends on definition and scope. The JWST team does state that things become more subjective. And that's okay! To me the art side is more manipulative, even misrepresentative, alterations, if serving only aesthetic purpose.

Agreed! I used that word, because of the subject of this thread. I don't like it either. We could use 'enhancing', like it is used in Ivo's table.

BTW: When are we going to continue filling up our table? I'm curious about the other modules, Super Structure, Denoise ...
But of course, I'm not sure if this isn't going to be a 2025 edition ...

Best regards, Dietmar.

[Mike in Rancho]

BTW: When are we going to continue filling up our table? I'm curious about the other modules, Super Structure, Denoise ...
But of course, I'm not sure if this isn't going to be a 2025 edition ...

Ha. Well Dietmar, 2024 still has a lot of time left in it.

After additional reading especially a couple more articles on Illuminated Universe, and of course the Sharpening experiments done earlier here, I'm wondering if I am loosening up a bit. Lowering my standards, even?

It's also dawning on me that these things are all quite mushy and thus hard to categorize, especially the more advanced and complicated tools like we find in ST (and elsewhere). So I guess I am good with Contrast, HDR, and Sharp all lumped together.

And while I now see the logic behind Ivo's table better, I still wonder if it might need to be sliced up more, or re-labeled? The center column and row don't even have names, just sort of "in the middle"?

And honestly, Contrast-HDR-Sharp seem to be "enhancing," now that we have discussed them a bit, don't they? Enhancing in a good way, of course. While the last column, Shrink and the others, strike me as going beyond enhancing. More...aesthetic manipulation and alteration. At least that's if we consider "enhancing" to be changes that reveal what is actually there, or increases acutance/perception.

As an example, I think we need a place to slot in tools like dynamic narrowband (Foraxx being one of them) combinations. Technically, that is fully intrinsic to the data, as no masks are used. Also, one can say (loosely) that the tool enhances or reveals what is there. But, the end result of that little pixel math trick between two channels is kind of a misrepresentation. Well beyond even just compositing different non-linear stretches. What do you think?

I have started thinking about denoise, re-reading the Features and Docs page, and also the Wikipedia page on NR. I don't really have an understanding of what's going on under the hood yet though. There seem to be numerous algorithms to choose from, and there was an inference that we are running into our old friend "blur" once again. Mabye.

SS has a lot of functions, so that one could be difficult. I imagine it is intrinsic, but I do believe hidden masks are created and utilized, even if they are automatic (i.e. linear luminance based). At first thought, and even if there's legitimate logic behind it, I am leaning towards it being more like Shrink and other aesthetic alteration than belonging in the enhancing category.