What to do about stars

[MartinB]

So the SGL star challenge seems to have stimulated some thoughts about the treatment of stars in astroimages.  So, is star suppression the work of the devil or can stars impair the visual impact of faint nebulosity.  Are the latest star suppression tools being used to take things too far or are they a true blessing.  Obviously there is no correct answer.  Discuss, compare and contrast...

[happy-kat]

Stars or rather no stars when I look at an image are a consideration of am I liking what I see it's not a right or wrong, when no stars is it crossing to art over form or illuminating the target for clarity. My aim is to enjoy what I do to the best of the data I have however I present it.

Edited January 16, 2023 by happy-kat

[Ratlet]

I prefer stars over no stars, I just think they always add something to an image.  They can provide context and can show how the structure is shaped by them (sometimes).

Ultimately though it's pretty much up to the imager.  There is no such thing as a real image, like this is how it should look.  They're stacked and sigma'd and stretched, so one more bit of wizardry isn't that big of a deal.  Heck, I'd argue that star suppression is no different than filters.

So long as what you are doing adds to the images, then fill your boots.

[Alien 13]

Definitely stars and ideally with JWT spikes, not a fan of the star size reduction software either as it produces fakes....

Alan

[tomato]

Being able to remove the stars has made my image processing so much easier, but I’ll always put them back in. These days I’m probably guilty of reducing them a bit too much, as a voice keeps telling me they should be tiny points of light, never discernible disks.

[The Lazy Astronomer]

Definitely with stars, but areas with a high density of them can look far too busy (in my opinion), and distract away from nebula structure and detail. Reducing them to push them into the background a bit, I think, benefits an image.

That, and, it's amazing how much noise is hidden by the inclusion of stars!

[ONIKKINEN]

Reduction, or rather not stretching them too far in the first place by separating the stars and nebulosity partway through processing, is a good thing IMO. Most types of images benefit from this, but not all. As an example i would give an image of M33 where it makes no sense to try and reduce stars by any significant amount since all the interesting detail is in the form of stars within M33 itself. (here i think even BlurXT fails to improve and rather makes it worse). I have a principle that bright stars should remain obviously the brightest points in an image, kind of like focal points for the observer seeing the image going "hey that thing has giant diffraction spikes and is big and bright, must be a big star". Generally an A is brighter than B in real life, so A has to be obviously brighter than B in the image kind of mentality should be kept in mind. An image has failed in my opinion when all the stars look roughly the same soulless blobs without any real colour (or the ability to differentiate the colour because of lacking saturation) to them and at that point star reduction has done no favours. There is also an argument that smaller stars are more real since they are actually point sources from our perspective, its just that blur added by the atmosphere, and the setup used smears them out to several pixels. Too much reduction is much worse than no reduction IMO.

1 hour ago, Ratlet said:

There is no such thing as a real image, like this is how it should look.  They're stacked and sigma'd and stretched, so one more bit of wizardry isn't that big of a deal.  Heck, I'd argue that star suppression is no different than filters.

I disagree with this statement by quite a bit, the basis for all astrophotography processing that is not specifically stated to be art should be realism and its not too difficult to stay within the lines. Stacking is no different than addition of signal over time the same way a longer single exposure is, its just a necessary step since we cant take 10 hour exposures easily from Earth. Stretching is also a necessity, since there exists no display that allows you to display display the brightness differences found in space (or human eyes, which are incapable of seeing those anyway). I think there is a general range of acceptable adjustments you can make for cosmetic reasons and its obvious when you step too far into fantasy from a realistic looking image (complete star murder one of them), but who gets to decide where the line is? And we go back into square one, since everyone has their own ideas of an acceptably touched up image so in an argument with 10 people there might be 11 different opinions. So also kind of agree actually, but not because of why you think so.

[Karl Perera]

I'm now removing stars early on in my workflow, stretching and working on the details of the image then putting the stars back in. It's improved my images but I wonder if I am overdoing the star reduction...

[ollypenrice]

What is natural?  The question is ambiguous. Does it mean What is natural in the physical universe?  or What is natural in the perceived universe? If the former then, as Steve says above, stars should not occupy more than a pixel each and the starless image is, by this definition, the most 'natural.'  If we are talking about the perceived universe, however, the stars will have dimension in accordance with their visual magnitude.

But... in the perceived universe, who is doing the perceiving? In the case of imaging, the answer is The camera.  What does the camera perceive? This:

Wow, that's a winner, eh?  £13,000 well spent!

So now we say, Well no, it's still natural if you stretch it...  And move the black point...  And use noise reduction as long as it still () looks natural.   And tweak the colour balance...  And remove gradients...   This argument is beginning to look a bit lame, don't you think?

Introducing a simplistic dichotomy like Science or art? does the conversation no favours because imaging does not have to be either, and because the use of art-like techniques can enhance the degree of objective information contained in the image. Take the image, above, of... erm...  👹 If you don't know, ask yourself how much information it contains.

In my image processing I am trying to extract from my data that which allows me to emphasize selected aspects of the view collected by the setup. And, surely, so is everyone else. Emphasis, by definition, cannot be applied to every aspect of the data, so the emphasis I lay upon the stars is my decision, one of dozens during the processing. There is no resemblance between emphasizing and inventing, inventing being acceptable in art but not in AP.

As a means of star control, removal and replacement has one huge advantage: it preserves, if done properly, the relative dimensions of the stars as captured, in a way that previous methods could not.

Olly

 

 

 

[CraigT82]

Star suppression is absolutely fine in my view. You’re simply mitigating a limitation of your data acquisition system, in the same way that noise reduction, sharpening, applying flats and darks etc is.

 

[MartinB]

Russ Croman, the creator of StarXterminator explains that his star removal is intended to allow stars to be processed separately to the rest of the image.  This is invaluable because the characteristics of stars are quite different to nebulosity.  How an image processor chooses to do this is up to them but the normal "rules" should apply, specifically, leave no jarring processing artefacts.  I like stars to "sit comfortably" in an image, I know what I mean by this but struggle to explain it!

[ollypenrice]

1 hour ago, MartinB said:

Russ Croman, the creator of StarXterminator explains that his star removal is intended to allow stars to be processed separately to the rest of the image.  This is invaluable because the characteristics of stars are quite different to nebulosity.  How an image processor chooses to do this is up to them but the normal "rules" should apply, specifically, leave no jarring processing artefacts.  I like stars to "sit comfortably" in an image, I know what I mean by this but struggle to explain it!

You explained it perfectly. Stars shouldn't look 'stuck on' and, sometimes, a degree of artifice is required to make them look natural.

And round we go again!!

lly

Edited January 17, 2023 by ollypenrice
typo

[teoria_del_big_bang]

The above two posts describe how I would hope to acheve and far better and far more concise than I would have explained.

18 hours ago, tomato said:

These days I’m probably guilty of reducing them a bit too much, as a voice keeps telling me they should be tiny points of light, never discernible disks.

However, I am also a bit the same as  tomato depending on the target. If there are so many stars and the nebula is maybe feint I would reduce the stars more than normal to allow the nebula to stand out more., but if not too many stars in the image I may not reduce them so much.
I do not really like a star;ess image as a final image but do use starless images during the processing.

But I guess as implied above in some posts none of these images are natural to a human eye at all and most of all if it is your image then I personally think it is up to you what you do,.
If everybody took astro images with the target the same fov, same view of the stars, same colours in the nebula then why bother showing it to everybody on SGL, we would have seen it all before.

Steve

Edited January 17, 2023 by teoria_del_big_bang

[ollypenrice]

^^

Dead right.

Olly

[ONIKKINEN]

Okay i am definitely weird then, my idea of a perfect broadband image is a boring one that doesnt stick out of the crowd too much, almost like a snapshot from some professional sky survey, and im okay with having seen that image posted by someone else before me (and after me).

I do think an image can be "real", especially if compared to an image that is comparatively not real, i choose to die on this hill!

So that the thread is not further derailed from the subject of stars to summarize my rant(s): Stars should not disappear, or cover the subject entirely and they should be sufficiently coloured to easily see that there are different types of stars.

[ollypenrice]

52 minutes ago, ONIKKINEN said:

Okay i am definitely weird then, my idea of a perfect broadband image is a boring one that doesnt stick out of the crowd too much, almost like a snapshot from some professional sky survey, and im okay with having seen that image posted by someone else before me (and after me).

I do think an image can be "real", especially if compared to an image that is comparatively not real, i choose to die on this hill!

So that the thread is not further derailed from the subject of stars to summarize my rant(s): Stars should not disappear, or cover the subject entirely and they should be sufficiently coloured to easily see that there are different types of stars.

There is a problem here!  An image which doesn't stick out of the crowd too much does stick out of the crowd - because there is not much wrong with it.  AP being very difficult, not many images, mine certainly included, don't have quite a lot wrong with them. Producing a clean, simple, unforced, slightly understated-looking image is, in fact, phenomenally difficult.  But it looks like we agree that it's something to aim for. 

Olly

[tomato]

It is a bit off topic but I think I get at least the concept of the ‘just right’ Astro image. For example, some amateur APODs are breathtakingly good, but I’m less enthusiastic about others with (IMHO) garish colours and over done detail. 
It is indeed incredibly hard to get it just right, to date in 8 years of trying I think I may have got somewhere close to the sweet spot a couple of times but that’s all. If we could process the detail and colour in images scientifically ‘by numbers’ I would probably embrace it but despite some recent advances I don’t think auto processing is here yet.

[Elp]

When you're imaging at f2/f3 widefield, the sheer amount of stars you pick up can be overwhelming, I think in this instance some minimisation is acceptable (to reduce/remove the faintest ones) to put more emphasis on the target. If it's a cluster then leave it as is.

Edited January 18, 2023 by Elp

[teoria_del_big_bang]

17 hours ago, tomato said:

If we could process the detail and colour in images scientifically ‘by numbers’ I would probably embrace it but despite some recent advances I don’t think auto processing is here yet.

I know some people really hate the processing side of this hobby, some love doing it and will willingly take on the task of having a go at processing somebodies  data to help them out to show what can be done with their data.
I am in between at the moment, I used to hate processing because I just couldn't do it like some others are so good at but with more practice I am jumping the fence a bit and getting to like it.
But really do we want auto post processing ?
I am all for auto pre processing, as that is purely mathematical to get the best results, and I guess sometime, not too far away in PI for sure there will be scripts to do all the post processing (there already is and actually does a fair job of it if that's all you want) but do we all want the same results from our data ? or at least nearly the same as some people will always have better gear than you, better darker skies, more time available and so on, but essentially they would all start to look very generic.
Or, do we still want that artistic side to it to interpret our data as we see fit ?

I am no authority on this by a long chalk but still maintain there are no natural, true to life astro images of any DSO taken from Earth. 
Yes we all comment about others images and often say that's so natural looking, normally as you say when colours are not so garish and in your face , but even  LRGB, to my way of thinking, are not really natural, we could never see what we process images to show, so many stars, and colour.  NB images certainly not natural colours not even the stars unless we falsify the colours by colour calibration of some sort.
But we look at so many images we perceive what the proper colours are, or what they should be, and even whereby we expect many of the DSO's to be with a SHO pallet, totally unnatural, and who's to say we all expect to see the same colors.

I have not been on SGL that long but when I joined I remember seeing quite a lot of mono images, and to me starting out having seen all those wonderful HST images wondered what the heck was that all about, but really are they mot more natural images ?  I think we think maybe not as interesting and maybe that's why we do not see so many these days.

I don't really know what is right, just what I think , but I guess we are not all the same and want different things from what we are doing, and that probably how it should be 🙂 

Steve

Edited January 18, 2023 by teoria_del_big_bang

[ollypenrice]

^^Yesss!

Auto-processing and I'm out. Processing is the bit I enjoy, the enjoyment coming in two flavours. 1) I feel like an archaeologist combing through the data to find what it contains. This is thrilling. It's like observing when you patiently work away at what you can see in the eyepiece. 2) I like to make a picture. Note, a picture.  I don't mean an invisible window onto reality - because I don't believe that's possible anyway. I want my picture to be a picture.

In my stumbling forays into regular photography I want the same thing. What I'm looking at is one thing, the picture I make of it is another. I want it to say something about what I was looking at but I want it, also, to look like a picture. I never expect it to be what I'm looking at. The difference is subtle but it exists. This is a picture. It looks like a picture. It doesn't quite look like what I see when I stroll down the road and look in that direction and that's what I wanted. 

 

[Elp]

I look at it this way, the old art masters were not created with automation. Took time and manual effort to learn and hone, even something like a Van Gogh has a human element behind every single brush stroke. If it were easy, why even bother?

Edited January 18, 2023 by Elp

[ollypenrice]

1 minute ago, Elp said:

I look at it this way, the old art masters were not created with automation. If it were easy, why even bother?

That's right. The day will probably come when a camera can replicate precisely the intensity and colour produced by light falling on the eye. This will hardly make an old master. The masters see something in what they are looking at and paint not what is there, but what they see.

Olly

[ONIKKINEN]

I would say its only a matter of time when we have a tool like an all encompassing AI where the user needs little to no input to turn a linear file into a processed, sharp, colour accurate and aberration free image (EffortXTerminator maybe?).

Not saying its the next few years but 10-15 years? Maybe, the field of AI infused software moves so fast its really hard to tell where it goes next.

Also not a fan of something like this, but there is a market for it i think so only a matter of time it happens.

[Elp]

It already exists, search engine + space agency + target. Effortless pixel math.

 

[DaveS]

I think of the dictum of Ansell Adams, that the negative is the score, and the print is the performance. If we substitute negative for our linear file(s) then perhaps we can extend his darkroom skills / magic to our postprocessing. Maybe taking a "Group f/64" attitude.

Or maybe I'm talking cobblers. Won't be the first time 😂.