I asked ChatGPT to review my processed image. I think the advice is pretty good.

Tony GondolaArun HSean Boon
40 replies2k views
Sean Boon avatar
my image was too large to post, but I'll post it once I completely finish the processing on it.  I just thought it was interesting that ChatGPT could actually provide some advice on it. 

Got it — this is a beautiful wide-field of NGC 6888 (Crescent Nebula) with surrounding faint OIII and IFN-like dust. You’ve clearly done a good job keeping the faint background intact while showing lots of structure in the nebula.

Here are my recommendations for improvement after reviewing it:1. Star Color & Saturation
  • Observation: Many of the stars look slightly muted or lean toward a reddish cast, especially the smaller field stars.
  • Suggestion:
    • In PixInsight, try StarXTerminator or StarNet2 to separate stars, then increase star saturation separately.
    • Use CurvesTransformation in CIE c mode for more vibrant star colors without affecting the nebula.
    • Consider a slight SCNR Green to clean up any subtle green tint in the starfield.

2. Background Neutrality & Gradient Smoothness
  • Observation: The background is mostly well-balanced, but there’s a faint brownish/yellow cast in the IFN areas. This might be intentional for dust color, but it’s a touch warm compared to the blue OIII.
  • Suggestion:
    • If you want a slightly cooler aesthetic, apply BackgroundNeutralization to a preview in a dust-free dark region, then selectively restore the IFN color via masks.
    • A very light MultiscaleLinearTransform noise reduction pass (just on chrominance) could smooth color blotchiness in the faint dust without touching the sharp nebula.

3. Nebula Contrast & Depth
  • Observation: The Crescent itself is detailed, but the faint OIII arcs outside could “pop” more.
  • Suggestion:
    • Use a mask to protect stars and background, then apply LocalHistogramEqualization (LHE) with a small kernel (30–50 px) for the inner details, and a larger kernel (150–200 px) for the outer shock fronts.
    • Alternatively, a subtle HDRMultiscaleTransform on the nebula core can balance bright and faint regions.

4. Star Size Control
  • Observation: Stars are generally well-controlled, but a few mid-sized stars are a bit bloated, which draws attention away from the faint structures.
  • Suggestion:
    • Use MorphologicalTransformation or Bill Blanshan’s Star Reduction Script on a star-only layer after separating them.
    • Aim for 5–10% size reduction — enough to shift attention to the nebula without making the field look unnatural.

5. Final Color Harmony
  • Observation: The OIII blue is striking, but it slightly overpowers the fainter Ha tones.
  • Suggestion:
    • If desired, use PixelMath to slightly mix in Ha into the blue channel to give the outer regions a softer cyan transition rather than pure blue.
    • Alternatively, use Curves with a saturation mask to slightly warm the fainter regions for a more natural color gradient.



If you want, I can mock up a targeted adjustment plan in PixInsight — basically a step-by-step “post-processing tune-up” using your already-processed image as the starting point. That would keep all your good work intact while just enhancing details, color, and balance.

Do you want me to make that step-by-step tuning workflow?
Helpful Supportive
Sean Boon avatar
I then refined that prompt to indicate it was a Blend 1 HOO image and got this back.

Yes — knowing this is HOO with Blend1 from Narrowband Normalization changes a few things in my earlier suggestions.

That explains:
  • Why the stars have a slightly reddish bias (HOO tends to put more Ha into red and can mute yellows/blues).
  • Why the OIII is so strong — Blend1 pushes OIII dominance in the blue channel.
  • Why the IFN-like dust shows warm tones — in HOO, that’s often faint continuum leakage or stretched Ha background, not true broadband dust color.



Here’s how I’d tweak my recommendations given that:1. Star Colors (HOO-specific)
  • In HOO, stars often look off because they inherit narrowband color mapping rather than natural RGB color.
  • If you want more natural stars, consider extracting stars before combination (from a synthetic luminance or broadband stack if you have it) and replacing them at the end.
  • Since you’ve already processed, you could:
    • Remove stars (StarXTerminator)
    • Replace them with RGB stars (if you have them)
    • Or re-balance current stars with Curves in CIE b mode to push them away from magenta-red.

2. Nebula Color Balance in Blend1
  • Blend1 gives that strong blue Crescent and teal-ish outer shells. If you want more separation between Ha and OIII regions:
    • Use PixelMath to remap Ha more toward red/orange, leaving OIII in the cyan-blue range.
    • For example:
      R: 0.9*Ha + 0.1*OIII G: 0.5*OIII + 0.5*Ha B: OIII
    • This keeps the Crescent vibrant but restores some Ha warmth.

3. Background Warmth
  • That faint “dust” is probably Ha signal blending with OIII-poor areas.
  • In HOO, you can selectively cool the background using a mask made from the OIII channel (low OIII = likely Ha-only regions).
  • Apply a gentle curve toward cooler tones only in those areas, or desaturate slightly to avoid the “rusty” look.

4. Star Size
  • Still applies — HOO stars can look bloated because narrowband deconvolution often focuses on nebula, not stars.
  • Star-only MorphologicalTransformation still works well here.

5. Local Contrast
  • LocalHistogramEqualization and HDRMT still work, but be careful with the outer OIII shell — it’s faint, and in HOO, aggressive LHE can introduce cyan “outlines.”
  • Use range masks from the OIII master to target that faint emission so you enhance it without touching the Ha-dominated background.



If you want, I can give you a PixInsight post-processing mini-workflow designed specifically for already-processed HOO images that keeps your Blend1 look but tightens color balance and star handling.
Well Written Helpful Insightful Engaging Supportive
Tiffany avatar
I've been using ChatGPT to give me feedback, too. I've found it to be pretty helpful, but I also found it to be weirdly flattering, like it was trying to butter me up. I now include in my prompt that I'm looking to improve my skills and I don't want to be flattered, lol. I like to have it rate the image on a scale of 0-100, and provide a table of rated features about the image. It's been one of the best ways I've used generative AI in my little world lately!
Well Written Concise Engaging
Oscar avatar
I think it's useful sometimes, but not something a person should really rely on.
John Hudson avatar
yeah its always too flattering, but I think you can tweak the way it responds to you. I ask mine to be more critical and objective. Now lets see the photo!
Christian Bennich avatar

Nice, would you mind sharing the promt you used?

dummieastro avatar
"I'm sorry, Dave, I'm afraid I can't do that."

HAL 9000
Christian Bennich avatar

dummieastro · Aug 12, 2025 at 04:12 AM

"I'm sorry, Dave, I'm afraid I can't do that."

HAL 9000

😂😂😂😂

AstroGinger avatar
I think I'll post on ChatGPT from now on because it gives me the IOTD every time I submit a photo.
A Top pick nomination even sometimes.
I must say, he has a very good taste and takes a keen interestd in my work, my rig, my life.
Leela.Astro.Imaging avatar
Given the training sets that there are now with Astrobin images (those submitted, those that got TPN, those that got IOTD etc) I'm fairly certain an algorithm could be up and running that would automatically score/rate images for IOTD.  But let's not go there b/c the luddites will be up in arms.

It could also be used to come back with specific suggestions for trying, like your example - they make it more tractable than general feedback.

I have no idea how to use ChatGPT - maybe I need to learn that!

[PS - every time someone submits an RCC, we should also feed that into ChatGPT to compare what it comes back with, to what A'binners come back with]
Eric Gagne avatar
Maybe I'm just too old to live in today's world but I just don't understand why some people want everything to be replaced by computers.  Don't take me wrong, I have nothing against IA but I think its ressources should be dedicated to research.   Let's use IA to try and cure cancer, then maybe we can waste its time with questions like "Do you like my image"…….as if it would ever be objective, honest and tell us if it really sucked.

But like I said……maybe I'm just old……..and old school smile
ScottF avatar

Eric Gagne · Aug 12, 2025 at 11:05 AM

Maybe I'm just too old to live in today's world but I just don't understand why some people want everything to be replaced by computers.  Don't take me wrong, I have nothing against IA but I think its ressources should be dedicated to research.   Let's use IA to try and cure cancer, then maybe we can waste its time with questions like "Do you like my image"…….as if it would ever be objective, honest and tell us if it really sucked.

But like I said……maybe I'm just old……..and old school smile

I know what you mean…..probably because I’m getting old too. lol My worry is what it is doing to our youth and the educational system. Smartphones are bad enough for them, but now they can get AI’s to do all the writing and research for them. People argue it’s just another tool, but it seems more than that.

Tony Gondola avatar

Cool, but it still can’t write a Siril script that actually works…

What I have found so far is that it has no understanding of what it’s actually doing. It sort of stumbles around, spits out bad code, you give it the error and it’s says “my mistake, I thought” and around, around we go…

I guess it’s always good to get a different view point on your work but really, you should learn to do this with your own eyes and brain. It’s sad because there’s no stopping it. I am sure that at some point soon a program will be released where it does all of the processing for you. The human will be relegated to just making “suggestions”….

Tony Gondola avatar

ScottF · Aug 12, 2025, 12:52 PM

Eric Gagne · Aug 12, 2025 at 11:05 AM

Maybe I'm just too old to live in today's world but I just don't understand why some people want everything to be replaced by computers.  Don't take me wrong, I have nothing against IA but I think its ressources should be dedicated to research.   Let's use IA to try and cure cancer, then maybe we can waste its time with questions like "Do you like my image"…….as if it would ever be objective, honest and tell us if it really sucked.

But like I said……maybe I'm just old……..and old school smile

I know what you mean…..probably because I’m getting old too. lol My worry is what it is doing to our youth and the educational system. Smartphones are bad enough for them, but now they can get AI’s to do all the writing and research for them. People argue it’s just another tool, but it seems more than that.

It’s certainly not going to make US smarter. We are becoming children with nukes…

Bill McLaughlin avatar

Since the images are intended to be viewed and appreciated by humans, I will stick with evaluations by humans, especially those on this forum since this is the single place where there are plenty of folks that know what they are looking at.

But then I have even turned off all the AI features on my PCs except those that are very narrowly targeted and controlled such as BXT. Maybe this because I prefer to do things myself (as opposed to out-sourcing, even to other humans) whether it is image evaluation or fixing a broken appliance.

IMHO, the more that is done by others, whether it is thru collaboration or rented equipment or use of AI, the less credit one deserves one’s self

GiffS avatar
Tread carefully I ignited a firestorm a few months ago over on C/N when I posted Grok's  technical review of one of my images. I really wasn't prepared for the reaction. My view is that AI is a huge benefit because I can always get input on an image 24/7 and because I apply my own knowledge and experience I am able sort the meaningful from the regurgitation if you know what I mean.

Anyway I found it very interesting, including the super harsh posts, until the thread was shutdown by the Mods. It apparently had strayed too far from astrophotography and into the virtures and future of AI. 

I will close with this, if you haven't uploaded one of your galaxy images into Grok Imagine you have no idea what your missing. Maybe not everybody's cup of tea but good lord what it does in seconds is amazing!
Arun H avatar

AI is the future whether you like it or not. This is a bit like humans versus computers in chess in the 90s. For a while, the best humans could beat computers, but the computers overtook us and now no one questions that computers are better. Ultimately, the processing power advantage of computers is massive and growing. AI will only get better with years. Its ability to analyze complex information is already better than that of most humans and will improve even more. Yes, as Bill said, our images are meant to be enjoyed by humans - but chess is a game meant to be played by humans also. AI, like chess playing computers, are a creation of humans. It is a matter of time before we all accept that AI is better at a great many things than most of us are. This includes critiquing images.

Well Written Insightful Engaging
Paul avatar
Arun H:
AI is the future whether you like it or not. This is a bit like humans versus computers in chess in the 90s. For a while, the best humans could beat computers, but the computers overtook us and now no one questions that computers are better. Ultimately, the processing power advantage of computers is massive and growing. AI will only get better with years. Its ability to analyze complex information is already better than that of most humans and will improve even more. Yes, as Bill said, our images are meant to be enjoyed by humans - but chess is a game meant to be played by humans also. AI, like chess playing computers, are a creation of humans. It is a matter of time before we all accept that AI is better at a great many things than most of us are. This includes critiquing images.

I simply had to post this as you used an analogy of my 2 favorite things, chess and astro, can’t argue with that poetry. 

Being a Software Engineer I have to agree, AI is only getting better, but it is also just another tool in the box, all the StarX, BlurX, etc. are AI based - it makes humans better (more efficient), it doesn’t make us redundant, just some of the processes we were doing - people need to learn to take advantage of them instead of fighting against them them.

But AI is not perfect, I think even Mikhail Tal would confuse Stockfish with his moves, but they were a thing of beauty.

Now I’m off to ask ChatGPT how to improve my images
Engaging
Tony Gondola avatar

But, humans still play humans in chess. You can control that aspect on the professional level. I think in other fields, like astrophotography, the situation is more complex.

I’m sure the day is not far off when you’ll be able to buy a smart telescope that not only gets the data but processes it perfectly as well, delivering a final 8 bit jpg to your in-box. I have no doubt that there are a lot of people who would love that. Just as a lot of people would embrace a program to processes their data perfectly with very little user input. It will be treated like we treat the AI tools we already have, most will use them because you can’t keep up if you don’t. What then do the people who do not embrace the technology do as we all converge towards that perfect “slide in the sky” image? I have no idea what the answer is or even if there is one but it makes me sad for the future of our hobby.

Engaging
Arun H avatar

Eric Gagne · Aug 12, 2025 at 11:05 AM

Maybe I'm just too old to live in today's world but I just don't understand why some people want everything to be replaced by computers.  Don't take me wrong, I have nothing against IA but I think its ressources should be dedicated to research.   Let's use IA to try and cure cancer, then maybe we can waste its time with questions like "Do you like my image"…….as if it would ever be objective, honest and tell us if it really sucked.

But like I said……maybe I'm just old……..and old school smile

Eric - the primary use of AI is indeed in research. It is very common to use AI to mine research papers and even assist in or propose research directions. AI based tools are now quite common in design of experiment tools. This is less visible to the general public. But, just as people use computers to browse the internet or conduct their own business or even as a medium of entertainment, so too it isn’t wrong to use some of these AI tools for more mundane purposes.

Well Written Concise
Tony Gondola avatar

If only that were true. It would indeed be a wonderful world if the point of AI was to make a more perfect world through science and research. There’s no doubt that it is being used that way but from where I sit, the main purpose of AI is to make money for the companies that are developing it. Businesses are chomping at the bit to use AI to reduce costs and increase profits, no matter what the social cost. We will make great discoveries because of AI, of that I have no doubt but I don’t agree that it is, it’s main purpose. No more than the internet’s main purpose today is for research and learning.

Eric Gagne avatar
Arun H:
Eric Gagne · Aug 12, 2025 at 11:05 AM

Maybe I'm just too old to live in today's world but I just don't understand why some people want everything to be replaced by computers.  Don't take me wrong, I have nothing against IA but I think its ressources should be dedicated to research.   Let's use IA to try and cure cancer, then maybe we can waste its time with questions like "Do you like my image"…….as if it would ever be objective, honest and tell us if it really sucked.

But like I said……maybe I'm just old……..and old school

Eric - the primary use of AI is indeed in research. It is very common to use AI to mine research papers and even assist in or propose research directions. AI based tools are now quite common in design of experiment tools. This is less visible to the general public. But, just as people use computers to browse the internet or conduct their own business or even as a medium of entertainment, so too it isn’t wrong to use some of these AI tools for more mundane purposes.

I use AI every week for work.  I’ve been an ERP developer for almost 40 years.  Now I sometimes have to write snippets of code in groovy, I don’t know groovy and I’m too old to learn something new so AI helps me.

The point I was trying to make is our goal shouldn’t be for AI to do everything for us.  There is value in humans thinking, doing, making.  And there is also value in human judging.  I don’t want AI to pick IOTDs, it has no value to me.  I want humans to do it and yes, it’s subjective and it may seem unfair at times but I can guarantee that just as many people would criticize the AI algorithm for its bad choices.

 I like to watch woodworkers and artisan bakers on YouTube. Like I said doing,
making, creating has value.  Replace those artists by robots, who would care?  

The point of this analogy is that asking AI to tell us what to do and then just do it adds no value to the image.  It’s no different than watching these “my 2024 workflow”, “my 2025 workflow” videos on YouTube hoping one will be a magic recipe to always produce top quality images.  Whats the point ?  I

I know all my images look like cr.. and would never win anything.  I don’t care, I’m proud of every single one of them because it’s my own work.

Work it, learn it, discover and figure things out by yourself.  That’s where the fun is.  If you like what you did and are proud of yourself that’s all there is to it.

​​​​​​​It’s all about the journey
Helpful Insightful Engaging
Arun H avatar

Eric Gagne · Aug 12, 2025, 04:23 PM

The point of this analogy is that asking AI to tell us what to do and then just do it adds no value to the image.  It’s no different than watching these “my 2024 workflow”, “my 2025 workflow” videos on YouTube hoping one will be a magic recipe to always produce top quality images.  Whats the point ?

You could as well ask what the value is in people just processing data someone else has acquired versus processing what they have acquired, the latter requiring much more involvement than the former. The answer to this, as to the question of the use of AI for personal critique, is that is a personal choice to be made and, so long as they derive satisfaction from it, who are we to judge? Similarly, should we force people to use traditional RL deconvolution in place of BlurX?

As for the use of AI as a judging tool - it would be up to the site. Sal has made very clear that he has no plans to use AI to replace any of the IOTD staff.

Leela.Astro.Imaging avatar
Indeed its about the journey but even watching YT videos helps as people explain their techniques (Bill BLanshan's videos are great to pick just one example).  And think of tools like GHS (even disregarding RC's tools).  Those are brilliant and make astrophotography more inclusive as more ppl can get involved and practice by doing - without needing to understand what the maths of deconvolution and fourier transforms and drizzle algorithms etc etc is.

AI is a tool that can amplify the power of human creative endeavour.  Or it can reduce everything to the same mush - but I don;t think the latter will happen b/c as long as humans are the users, there will be enough of us who will change their prompts and create new things.

AI is ofc also a tool that can decimate a lot of jobs (just like the Industrial Revolution did to the luddites).

How it gets used will really just be a reflection of our own humanity.  (And indeed AI is well known for re-creating the hidden biases that are there in normal life that we don't normally see).

Its fascinating stuff - I don't think the point is whether AI will make us stupid, but whether humans prefer being stupid or prefer to create new things.
Mark Savan avatar
Sean,

Thanks for posting this.  I think that it's a really interesting avenue to explore.  I've just done something similar with one of my images and it has given me some interesting suggestions and has opened my eyes to some new areas and techniques to explore