Topaz = FAKE ASTROPHOTOS?! AI Tools in Astrophotography - HOW should we use them?

This is a really difficult topic to touch upon for me because I think it can be very controversial, and it's really open-ended.
I hope I don't cause flame wars, and that I'm being fair to Oscar.
Support me on Patreon! / cuivlazygeek
Support me by buying ANYTHING from Amazon after following the below link!!
amzn.to/3hTB5Ne
OPT Affiliate Link: bit.ly/2OIw6jH
RENEWED! High Point Scientific affiliate link: bit.ly/3lReu8R
Merch: / cuivthelazygeek
External shop link: cuivthelazygeek.myspreadshop....
Follow me!
My Instagram: / cuivlazygeek
My Facebook page: / cuivlazygeek
Order of the Lazy Geeks slogan: Christopher T
Merch Design: Radu Chelariu
#astrophotography
#astronomy

[@CuivTheLazyGeek]

This is a really difficult topic to touch upon for me because I think it can be very controversial, and it's really open-ended. I hope I don't cause flame wars, and that I'm being fair to Oscar and his beautiful image. Oscar's beautiful astrophotography gallery: www.oscarviteriphotography.com/ Support me by buying ANYTHING from Amazon after following the below link!! amzn.to/3hTB5Ne Support me on Patreon! www.patreon.com/cuivlazygeek OPT Affiliate Link: bit.ly/2OIw6jH

[@dankahraman354]

sorry I won't be patronizing Oscar's beautiful fakes!

[@Tim_the_Astronurd]

i feel the same way about blur exterminator

[@mastroitek]

I'm not agains the use of AI during editing, though that image looks...wrong... I have little experience in AP, but I clearly remember that after the first few months I started to understand how images should look like and that one is the complete opposite of that. Though in the end if he likes the result than great, I just look for something else in my astro pics

[@lwizzit]

It’s an eye of the beholder question. Think about color. When NASA revealed that beautiful picture of the Pillars of Creation, we all began applying the Hubble palette to our work. Is it an honest representation of the nebula? Not really. But it enhances detail and pleases the eye. With a few exceptions, most of our photos would be solid red being swamped with Ha. So we use other color palettes as an enhancement technique. Should we say now that people with OSCs using Bill Banchard’s tools are « introducing » false details when they are getting essentially the same results as those shot in mono? It’s a slippery slope and artistic license exists whenever we so much as pull up the curves tool in Pixinsight. In the end, if you like your image you have a great image.

[@williammauney1129]

When I first started I really focused on following every step, every process, etc. to get my image to look exactly like others. And I soon didn't want to continue the hobby. In the past few months however, I have a basic workflow for processing that gets an image that my family and friends like and has started conversations on getting others into the hobby. Even a few pictures on the wall. Every few weeks I will try to add a process that I may not have before to get just a little better. My shot of Rosette from this year is worlds better than it was last year. But not obsessing about it. I do try to get the colors 'right' and I am using the XT suite in my workflow (which some have said is 'cheating'). But I'm taking the pictures, others are enjoying the final results, and that's good enough for me. Next year's Rosette maybe even more amazing than this year's.Or it may not. Doesn't matter. I'm going to enjoy what I'm doing first and foremost

[@rudyamaya2532]

You said it well. The hobby is for me and for me to share. Just knowing that when I look at my photos, I can say "I did that, I shot that photo with my own equipment." The hobby of astrophotography is so that I can look at a photo of a nebula, star cluster or galaxy and just be amazed of our universe and wonder what else is out there. After all isn't that what astronomy is about. We just happen to be taking pictures of these object. I too have stopped comparing my photos with others. I am just interested in revealing as much detail as possible.

[@lewisgreiner1977]

Absolutely a great and honest conversation. Thank you so much! The subject is beyond complicated. I remember when digital photography first emerged -- there were these guys walking the streets of Denver, Colorado trying to sell very, very over-sharpened landscape shots, that to some people looked fantastic. As a long-time (regular) photographer I knew they were "bogus" (whatever that even means) and therefore to me were unappealing. But other consumers loved them. And we've all seen those over-colored shots of wildflowers (sure the flowers are purple, but the stream seems to have a purple tinge too!). Frankly, nebulae are similar. What color is the Rosette, anyway? My most recent shot (a nice rendering mostly because the camera, a ZWOASI2400MC Pro OSC one, is really fantastic. But in final processing in PS I certainly turned down the red to what I felt was a more pleasing level. No matter how you look at it, that's art (maybe more like personal preference), not science. It printed up great 20X24 on metal, so I gave some away as gifts. I included all the data info (including the Topaz Labs deNoise AI step) -- I'm pretty sure the receivers of the image didn't give a damn.

[@CuivTheLazyGeek]

Thanks so much Lewis! It's very interesting to compare to landscape photography! I always look at landscape shots thinking the photograph captured what they felt rather than what they saw :-)

[@JoMoJack]

"I don't care, I just want to have fun taking pictures of stars and nebulae and processing them [and just like,] for my own pleasure." I think that sums it up perfectly. Each one of us is going to have our own motivation for astrophotography. We can do it for ourselves, we can do it for likes or for follows, we can do it for astrophotometry... there really will be no end to the justifications. I personally feel that it does get to be disingenuous if details about the processing are omitted in such a way as to represent them as natural. Then again, if that image isn't being used for proper data collection and analysis, does that really matter? If we were to hang that image on a wall are we being dishonest if there isn't a paragraph below it of all data acquisition details as well as all processing details? I don't think so, but I do think that as the source for these images we do have a duty to disclose that some image details may not be true to how it actually exists in the universe.

[@CuivTheLazyGeek]

Where the line is is really fluid and difficult :-) 😭

[@DylanODonnell]

I agree and have been comparing the two myself.. thanks for highlighting these examples.

[@CuivTheLazyGeek]

Thanks Dylan! As I was making and editing video, my mind kept coming back to that comet video you made a few years back with the details in the tail :-) That one was really egregious though!

[@jPaulofe]

Great video and topic Cuiv! In my couple of years on the hobby, I always try to capture the “real deal”, whatever that means, but trying not to introduce false colors nor too much processing, but I agree that some tools like the ones you describe (that I don’t own), might be helpful to give an extra “touch” to the final image and to reduce visual problems that are very difficult to cut otherwise. Well… I think that it’s a balance, but this is an hobby and not a contest after all, isn’t it? 😅

[@setnes]

Thanks so much for talking about these issues. I think what we have in astrophotography is something similar to Barbie Doll syndrome. We see amazing images and feel bad about our own. This leads us down the path to use the latest and greatest software tricks to try to fill in the gaps. It's hard to know where to draw the line between art and accuracy. I guess it's all about personal choice. I don't like starless images because of how much false data exists, but I certainly use Starnet as part of processing. I just prefer to put the stars back.

[@fazergazer]

I do stereoscopic photography as a hobby, as well as astrophotography. I like to set images with different processing side-by-side and then observe them as stereo pairs (left eye and right eye images). This allows me to spot differences instantly. 🎉 another way is to blink a stack.

[@PaulymanAstro]

Fantastic work Cuiv, very level and unbiased approach to the topic. I remember when Topaz first started making itself known in astrophotography. A particular imager won quite a few APODs and it was apparent something wasn’t right, quite a few people I recall outright felt they were simply photoshopping in Hubble imagery. But I had played with Topaz a bit and instantly recognised the tell tale “turned it up to 11” that is exactly as you describe here. I don’t go as far as to say it is fake data. I shave always felt it is precisely the same kind of issue that deconvolution has, sudden changes in contrast particularly in darker regions like dark dust clouds seem to get sharpened. I was on the lowest ranks of the Astrobin IOTD committee at the time and we were aware of and carefully looking out for Topaz artifacts, I assume that still happens, I haven’t been on the committee for a few years though. Very interesting video, we all seem to creating AI videos at the moment 😂, thankfully they are all very different and unique takes on it.

[@CuivTheLazyGeek]

Thanks so much, that means a lot - it's really tough putting out such a video because I know it will feel like I'm attacking others, but I tried to keep it as objective as possible!

[@barrymak421]

I see your point, but at the same time, I don't think anyone's work should be looked down on for using a certain processing tool. It's that that photographers picture, and they should be able to process to what their eye likes, and not have someone come after them saying oh this is a bad image because you used a tool that I don't like. Just my 2 cents though, and it has been a topic coming up more, so good for you for doing a video on it.

[@trevorgreen2232]

Thanks Cuiv. Another great presentation and this gives a great topic for conversation with others especially the AI used to create the background in Starnett / Star Exterminator once the stars are removed.

[@hexasquid]

Wooow, this is such a differentiated and well thought-out video!! It helped me a lot in figuring out my own opinion on this matter, thanks!

[@pamelawhitfield4570]

A very balanced discussion. 👍 I’m still a fan of the old non-AI Topaz Labs tools as they gave lots of controls for manual adjustments in Denoise, Adjust and Detail. I’m not a fan of the AI Topaz versions though - too many artefacts for my taste. I do appreciate the more targeted approaches taken in the RCAstro tools and use them regularly - although not turned up to 11!

[@CuivTheLazyGeek]

Thanks Pamela!

[@DSOImager]

A difficult topic. I think in regards to color there is more latitude. When we talk about natural colors.. we're really talking about how our own eyes and brain view them. Someone who is color blind is going to see the colors differently than someone who isn't... so how can we say what color is accurate :) The generating changes in the structure of the objects.. that's a different level in my opinion. I agree, it should be disclosed which AI is being used. From an art perspective I have no problem at all with any of it. Where I think there is risk is that stuff that isn't real is passed off as real. People not familiar will believe it's real.. and then when it's discovered to not be real.. it plants that seed of doubt.. and then real stuff gets dismissed as fake. For example, I was floored to see people calling out JWST images as fake. Before you know it, people will start to believe the Earth is flat!

[@CuivTheLazyGeek]

Wait, the Earth isn't flat?! :D Yeah completely agree with you! People have also been calling the Hubble images fake for ages...

[@ACKitsBilltheCAT]

As a color deficient astrophotographer, I resemble this remark. I see red and green, but not nearly as well as the normal folks... makes it extra interesting! LOL

[@joshuacarter3478]

@@ACKitsBilltheCAT Same here! Red/green is difficult for me. When someone says this image has a bit of green or red tint, I never am able to see it. I process my images to my liking and hope for the best.

[@skywatcherca]

Mr. Cuiv: all that I ask for, all that I expect is honesty - when I watch these reviews. YOU Sir, Mr. Cuiv are one of the ONLY honest reviewers! I'm not looking for perfection, we all make mistakes, and we all have our biases, but at least you try to be honest. As well, I don't care if you receive 'free' equipment from suppliers - that's fine, as long as you keep your reviews HONEST. I know of one fellow, a STOOGE: he just reviews as the manufacturer wants: his reviews are garbage and total 'crap.' He's a stooge, a clown - he does not help anyone (but himself), in fact, he does a mis/dis-justice to the viewer who may depend upon these reviews. In any case, we are not related, I have never met you, and I am (situated) on the opposite side of the planet - so with all of that having been said, your reviews hold value. Any/all manufacturers of astronomy equipment (and related equipment) should send you equipment for HONEST reviews: alias, you can bet that manufacturers who do not want an honest review, will avoid you like the plague. Keep up the great work. I don't know how to (further) promote you for your honesty: I am a subscriber. Thank you Mr. Cuiv.

[@komr323]

Haha I always say the same to my friend. I don’t want to always think what is true and what’s not when I watch youtube vid.

[@rcpattaya230]

This is truly an interesting conversation. But most important to remember is that this is a hobby. I think 99% of us are no astronomy scientists, we are just hobbiests. Our pictures are not being used for scientific studies, they are for our own satisfaction. I enjoy preparing my astro night, shooting my pictures, stacking and processing and that's it. I'm not sending my pictures to any institution for verification, I don't care whether it's real or not, I'm enjoying my hobby. If it looks "real", I'm happy. If it doesn't, I'm still happy. So, let's not over-judge our results. Only few of us can afford to spend the amounts to purchase equipment capable of producing near scientific data. Most of us are on a budget and just enjoy what we have. Who cares whether it is 100% correct or not. I don't. When someone creates a monkey in a nebula, we all know it's not true. 😉

[@CuivTheLazyGeek]

Hey, but there IS a monkey nebula! Have we been lied to? ;)

[@LogansAstro]

Very interesting topic. I've noticed opinions vary widely on what is considered a "good processing tool" and what is considered "cheating". I've even seen using the HDRMulltiscaleTransform tool described as "cheating" but other similar tools considered acceptable which I've never quite understood the rationale behind that. With regards to StarXterminator making up what it thinks lies behind the star - perhaps it's not so much of a worry so long as you put the star back and cover that area up again🤣.

[@CuivTheLazyGeek]

Also the comments on here are very interesting! It's very interesting to see all the opinions!!

[@ysl109812]

Great discussion, one that I have asked myself often. An amateur astrophotographer is of course always free to do whatever he/she wants. But in the case of the M82, though, I think it is relatively apparent to astrophotographers, who choose to use a camera instead of a paintbrush as a tool, the image would not be considered a great representation of M82. With Hubble's image as a reference, we do know what, and even how, artificial details were added to the image. Nevertheless, I believe, the line of what is considered "realistic" is a fluid one and it depends on the precision of our tools and our understanding of the object we photograph. The more complicated issue that I have been pondering lately is the one on colors, especially on RGB colors. With the newly introduced tool SPCC in PixInsight, we are now able to largely color-correct our images to be more realistic than ever, using Gaia data as a reference. So we do face a decision of whether to follow the guidelines closely, even if it is not artistically pleasing. In PixInsight's SPCC documentation, there are very specific "best practices" guidelines, which I personally find somehow debatable. I think that can be a very interesting discussion for a future video.

[@CuivTheLazyGeek]

It's indeed very fluid! SPCC is another can of worms... I'll need to get to it at some point :-)

[@dankahraman354]

Cuiv your honesty is appreciated...let's keep it real by relying on tools that bring out what is there not have AI make guesses for us.

[@fmrc69]

I see this in pretty much every over-processed M42 Orion nebula image posted in the last year or so.....insane cloud detail that just isn't there

[@cigarnationwarriors]

I’ve been a backyard visual observer for almost 70 years! Along comes the Seestar S50 and suddenly I can photograph those faint, grey fuzzies I see in the eyepiece. Any enhancement I might do is for my personal pleasure. Noise Reduction and maintaining Sharpness high on my list🔭👍

[@tubedude54]

I like looking at images that 'pop' when you see them. One color nebula may be more of what the human eye would see but our eyes are very limited as far as 'seeing' the various different wavelengths that are there. Seeing an image that is just 'red' to me is like looking at it in B/W. It's just boring. There is so much more to an image than the visible spectrum! Colors show variations in fine detail to some degree and make an image so much more appealing to the viewer!

[@lukomatico]

Very interesting video mate! Very fair points you raise! - I've been considering looking at Topaz and BXT/NXT head-to-head for a while too but I'm still waiting for my flame resistant suit to arrive in the mail first! :-D Such a divisive subject at times!

[@CuivTheLazyGeek]

Hahaha yes, you need a very very strong flame resistant suit :D

[@stuartwilson5772]

Excellent video and interesting topic as ever Cuiv. One of the few advantages of operating at the budget end of our hobby is producing low resolution images. My two astro cameras have 8.3 and 20 megapixels respectively. My file sizes and consequent processing times definitely save me a lot of time and I'm not sure I'd recognise fine detail if I saw it 😀

[@CuivTheLazyGeek]

Cameras like the 533 are great for that :D

[@stuartwilson5772]

@@CuivTheLazyGeek Recently I've mostly used my Player One Uranus C (IMX585) which is a kind of widescreen 533. Great little camera and here in Scotland at least cooling is not an issue!

[@starfleetau]

Would be nice to see this revisited with the changes to Topaz the splitting of denoise and sharpening etc. Great convo on this it's a hard one.

[@newforestobservatory9322]

Great topic - and very well covered. I would think that somewhere close to 100% of the people looking at this are not doing astrophotography for scientific reasons - therefore I am saying we are all in it for the "pretty pictures". That being the case what do "I" think is acceptable and not acceptable? I was going to say that you can do ANYTHING you want to your own data, so long as you don't add false data. I understand what you are saying about the noise reducer effectively adding false data, but in that particular instance, as we are not using the image for science, then I personally don't care at all. Then I thought that I use Noel Carboni's StarSpikes 4 whenever I think an image needs a little more punch - so the idea of not adding false data went completely out the window :) I think the only place where I would really draw the line in our attempt to make the "best" pretty picture is that you don't mix in professional data with your own (which is very easy to do of course) and then call it your own work. I would like to pick a very famous example of a great astroimage to further add to the conversation. There is a VERY famous M31 image made up from a large number of frames (a big mosaic giving VERY high resolution) and it looks GREAT. You see this particular image EVERYWHERE - and it is used as a "standard" by the astrophotographer community. But have you ever had a REALLY close look at the image? If you have you will have noticed that the overall SHAPE of M31 looks a little bit odd. And it IS a little bit odd because all the separate frames have not been added together correctly. Try overlaying your SINGLE frame image of M31 onto the top of the famous one and see what you get. However - it's still a great pic - and that's all the public are interested in. I'll leave it there :)

[@douglasfleming1]

Some philosophy this time round. Love it! Disclosure is a must, I agree.

[@scottrk4930]

This is the “can of worms” that needs to be opened . I’m sure we would all enjoy and benefit from some informal discussions from the YT Astro Experts . Maybe something you could organize ? BTW excellent image you took in your examples ./srk

[@PeterK6502]

I want my photos to be as close to real as possible. By 'real,' I mean what our eyes could see if they were sensitive enough. Therefore, I don't use AI-powered tools, fake colors like the Hubble palette, or star removal tools like StarNet. The problem is that many people use them, which creates a false sense of how images should look. For a long time, I did not understand what was wrong with my telescope and camera because I did not get the results I saw online.

[@jonathanclark6337]

Cuiv, as usual a very interesting and thought-provoking video - thanks! I'd just like to make some comments that might help some viewers. My research has involved AI for a number of years, and I've also taught it. So I'm fully aware (and this is a warning to AI novices) that with AI (as with humans!) it is very much garbage in garbage out. Firstly your input data needs to be good and as large as possible (the amount of data, eg. relating to the number of good subs, not the pixel resolution of the images!). Secondly, it is safer to use the AI to only slightly improve models/images. Otherwise (as you so clearly show) artefacts can appear. I have used the Topaz AI suite for about 2 years now, and I mostly use Gigapixel AI, which increases the resolution 'intelligently'. I think it is very useful, BUT I only ever create (only) 2x the original resolution AND I compare with a Hubble image to see if new detail prodeced by the AI engine is showing real detail or fake. If it appears fake, I discard the AI-enhanced model. Also, I only ever use the AI engine when I have done all the usual processing with PixInsight and Photoshop, that is, as a last finishing off process, and even then, with caution , when processing astronomical images. Very often, though, the AI does show detail that is in the Hubble images and not obvious earlier. The Topax AI tools are also very good to enhance conventional photographs.

[@dumpydalekobservatory]

Very interesting Cuiv I just use the tools I have which is Deep sky stacker, SIRIL, Starnett, Photoshop & Topaz denoise, I hope I don't introduce detail that isn't there in my images as all I use Topaz for is to remove the noise & that's it, I will switch to Pixinsight one day but I've got jobs to do with the observatory to get finished first. I do see images online that have shown insane amounts of detail & I do wonder sometimes on how they managed it so maybe that's the reason who knows.

[@CuivTheLazyGeek]

It's possible indeed that those insane details could come from Topaz

[@sachastayswi]

Hi Cuiv thank you for featuring my image (though I don't remember us discussing 🤔). Certainly I am not an experienced astrophotographer (started about 7 months ago) and I took this image with an OTA from a Dobsonian telescope. Wish I knew how to use Pixinsight since blurXterminator is not available for Photoshop yet. Thanks! -Oscar

[@CuivTheLazyGeek]

Hello Oscar! We interacted in the comments - sorry I hadn't realized I was on my personal account at the time, rather than my Cuiv persona! Starting 7 months ago is truly amazing in terms of what you can do already, and again your pic is really good :) Concerning BXT, my understanding is that it's technically difficult, and Russ has no plans to make a PS version :(

[@sachastayswi]

@@CuivTheLazyGeek hi Cuiv. No worries! I'm a total noob on this and I have learned so much from channels such as yours. I appreciate the thoughtful and unbiased review. I am also one of those that pulls the SHO palette on one color cameras. I just don't have the time to commit to multiple filters over several nights as I see patients during the day 😅🥲 I will learn Pixinsight. That is a goal 😉

[@FrootyRecords]

I used Topaz a lot to make up for my less than perfect data!. It is a great tool but I agree it can be very aggressive if not toned down significantly with the recover detail slider as you said Cuiv.. Now I'm am trying to get my data better in the first place so I don't have to rely on it so much. I have star exterminator and noise exterminator and will definitely be getting blur exterminator after seeing the excellent and arguably more faithful results it produces whilst still doing the job you wanted it for. Everyone should have those 3 tools at their disposal if you can afford them in my opinion.

[@rudebuddha4895]

Great subject for discussions. IMO, since from the get-go astrophotographers are using a camera that would be the equivalent of humans having nocturnally tuned eyes possibly the size of basketballs that are capable of unflinchingly looking for hours at one single object and accumulate light photons to create an equivalent "stacked" image in our brain, which is completely impossible and unnatural, it's all subjective interpretation from there. However I personally try to keep it within the realm of the data collected and not let the colors become cartoonish while maintaining a compelling image to greater or lessor extents.

[@CuivTheLazyGeek]

Good way of looking at it!! I like this approach

[@OtherWorldExplorers]

It has to be the compression of KZbin When you were comparing blur exterminator to topaz. To me on my computer at max settings I kept seeing solarization in the light blues. And in abuela City I seen a lot of artifacting around the edges and fringes. So it has to be KZbin.

[@gregerianne3880]

Wait, you DON'T have telescopes for eyes, Cuiv?! 😛 This was a great video. I remember when I first started doing astrophotography one of my first questions (that is still unanswered, really) was, "What is this SUPPOSED to look like?" That's why I gravitated to PixInsight. I thought it was less likely that I'd stumble into too much 'art' by doing things just to dress up an image to look better for people who see my images (and don't run away laughing or screaming). Difficult topic! I liked your approach.

[@CuivTheLazyGeek]

Thanks Greg!! And technically my eyes are telescopes (pretty crappy scopes though, but variable aperture) paired with an amazing sensor in terms of dynamic range, but the exposure time is really limited, sigh :p

[@siegfriednoet]

Good question and really hard to give an answer Quiv, depends on what you want to bring, just a beautiful piece of art or really science based images. Lets hold it on this, let everyone decide for himself if he/she wants Ai to make his/her image look different

[@CuivTheLazyGeek]

Exactly, super hard to give an answer, although I think it's important to be aware of it

[@taurus7228]

Hi. Very right video and at the right time for me. I have shifted to new software for the processing part of my astrophotography hobby. I tried re-processing many of my old images and I just cannot believe I took those. That truly made me question myself about some of the new processing softwares. I wonder if those AI based softwares are not adding fake details or adding data from other sources. 😅 Thank you

[@CuivTheLazyGeek]

Always good to be aware of it!

[@Neanderthal75]

Well, I'm still a beginner after 2 years, but lately the weather and endless clouds at my location put me on a place, where I am just happy if I just get a smudge in the middle of my framing lol! I'm not using any sharpening tool, beside noisexterminator, but sometimes not even that. I just get the noise down with various technics, but honestly I leave some if it's in the details of a nebula or galaxy. My fear is that some of that noise might not be noise but part of the detail and noise tools can fade out and remove in a favor of something that looks like water painted.

[@nikanj]

On a side note - The cigar galaxy image doesn't look right to be because of the star size. Point sources of light create a diffraction pattern called an "airy disk." The size of the airy disk is inversely proportional to the aperture size. The aperture size is also proportional to the maximum achievable resolution. To resolve so much detail in the galaxy but still have large stars is inconsistent.

[@ralforzel6818]

Hi, thank you very much for the very detailed consideration of the current topic of AI. As you described, it is a very difficult topic. But who wants to stand up and say what is right and what is wrong? Can the hobby photographer, you, Nasa with Hubble and JWST? Maybe the official institutions with their high-tech equipment. But at the end of the day, it's a matter of taste - what you like. A few days ago, I had NGC2024 in front of the lens and processed it afterwards. Apart from partially wrong colours, or other small errors, I was very proud of the result, as it showed me structures and details in this nebula that I had not expected. And that with "BlurXTerminator" and "UnsharpMask". A good friend of mine, who has many years of experience in astrophotography and whom I respect very much, did not like this at all. He then created something with the same basic data that looks a little more "washed out". What is right now, what is wrong? Nothing: "Beauty is in the eye of the beholder".

[@donlindgren109]

My rule is “Do whatever you want, but don’t pretend that you didn’t do it.” Astrobin has plenty of fields for disclosure. Everyone has their own style/preferences.

[@fazergazer]

As a wee child, my nephew thought that stars all had points. That’s usually due to astigmatism. Some folks hate diffraction spikes from spider vanes and the like, others like them. 😂

[@RazorChrist]

It's a very interesting topic for discussion for sure. I try to approach my own photos with roughly 70/30 percent, where 70% is trying to keep it true to the scientific raw image, and 30% making art out of it. I do use Topaz Denoise AI sometimes, but I've never really thought about "is it creating data that isn't actually there?" But another thing I've started doing recently is adding star spikes to my stars. I think it looks pretty, but it's still adding data that isn't actually there. It's a tough call to be sure, and every astrophotographer has to make up their own mind on how much or little to use in the way of AI tools.

[@CuivTheLazyGeek]

Very good point on the star spikes Razell! You can also get them natively with just elastic bands in front of your aperture :p

[@xpanderxsert2613]

Awesome. I really like your objective and transparent approach. Thanks to your video which answer my question "buy Topaz Package?". The answer is no! I'm will follow the scientific way! Thanks!

[@nikaxstrophotography]

If I use Topaz i make sure i use it on a different layer and the image is starless and let it do its "magic" then I will use a layer mask and erase any of those weird artifacts to ensure a more natural and accurate representation. Topaz does work well to get rid of excess noise and colour but its use needs to be toned down and minimalised. Having now started using Pixinsight I use topaz even less as blurxterminator and noisexterminator does its job fairly well

[@lordgeorgemaster]

I have been using topaz denoise, after comparing the pre and post pass you could say details become apparent that otherwise wouldn’t be so. That being said, it’s a typical scenario where over-processing destroys the original signal too far. I personally like the technology when used in modesty.

[@AlphazerstoererX]

I use TopazAi for about a year, my recommendation is to use the Raw or low light mode, put sharpening to 1 and noise reduction to 1 - 10. With this settings I get good results for the most cases. And for low noise Images I don't use it.

[@Zealor365]

"Artifact Generation" is found in most processing software. The software does not know what is there, and what is not there. The details you see generated are an exaggeration of the 'Tonal Contrast' of the image. It's perfectly acceptable to leave or remove based on the will of the artist who processes the image. Alas, there is no right or wrong...only do or do not.

[@neilsmith5883]

I myself like to keep my images as natural as possible. Only processing I do is the usual basic photoshop routine of selective curves and level balancing. One shot colour camera like the 2600 mc will provide the most natural image you can get. One of the reasons I haven’t gone down the mono route. I find mono images are about the science of the subject rather than what it actually looks like which I’m just not interested in. Though I love black and white.

[@jamesmcvicar8414]

In the end it depends on what kind of image you are aiming for and the uses to which you intend the image to be put. If you are trying to analyse deep structure in a nebula for scientific purposes, it would probably be unwise to apply a broad-brush denoising tool, such as Topaz DeNoise, to it before doing so. Similarly, it would be unwise to apply a deconvolution tool to an image if you were intending to use it for magnitude estimations (e.g. in variable star work). On the other hand, if you are trying to create an aesthetically pleasing photograph, why not use whatever tools you have to hand to produce what you judge to be the best image? So it's "horses for courses".

[@koppervik1]

I introduced Topaz denoise to my processing workflow (Photoshop) a few months ago, and I am not turning back. Having said that, while I initially frequently 'overdid'/pulled the slider way to far to the right, I now pull the denoise slider as far to the left as possible to have a good effect on noise, while most of the time setting the sharpening to 0. That way, I don't think I introduce non existent details, but I still get a good noise reduction, better than I am able to with the built in tools in PS. But I really don't see a difference in AI based tools and other tools applying statistics in processing. In both cases its educated guesssing. And really, as soon as you create a false color pallette, or you start playing with curves, you have already deviated from 'true science'. But as you said; why care when all we want is to create beautiful images? After you made a video with my Tadpoles and Flaming star image, I did go out and image the same again for about a total of 11 hours integration. I have reprocessed the same image several times. With 11 hours integration, noise was easy to handle, and in one reprocessing run, I restricted myself to not touch curves, saturation etc. Just levels and noise control to keep it as natural as possible. This version ended up as the one I was most satisfied with! 😊

[@CuivTheLazyGeek]

Great summary! I stopped using Topaz after using it on an image and seeing that no matter what I was doing and what settings I was using, some details that don't exist were introduced - it was pixel peeping though, but it just didn't work for me personally. I'm part of the crowd who things false color palettes are absolutely true science and just a tool to do science (Dr Becky has a great video on the topic called "is the color in space images "real"). I love to see how staying the truest to the original image is what you ended up liking the most :-)

[@koppervik1]

@@CuivTheLazyGeek But you are an advocate of no pixel peeping! 😄

[@captainfruitbatify]

I have the Topaz suite but have found it tends to go completely overboard on denoise and particularly on sharpening if you're not careful. I suspect that its AI has not had much, or any, training on astrophotography images. That may be fair enough, given it is a more general purpose product. I don't use Topaz Sharpen AI on AP images at all because it really does make a lot of stuff up. I do use Topaz DeNoise, but only on starless images, and only in Low Light mode, which seems to give the subtlest, most realistic (least "made up") results.

[@CuivTheLazyGeek]

Yeah Topaz doesn't seem to have been trained on any deep space astrophotography pics... Since BXT and NXT have become available, they really are what I use, but that's just me :-)

[@TedDobosz]

FYI I have contacted the Topaz team and asked them to add astro images to the AI training. I indicated that there is a significant number of non Pixisight users who might be interested in the Topaz route rather Pixinsight.

[@crm114.]

There are so many variables in creating an image there’s always going to be a high degree of artistic interpretation. My images are not going to be used for scientific reseach so I really don’t care that much. I do try and and be careful with all image enhancing tools so the final result doesn’t look over processed.

[@CuivTheLazyGeek]

In the end it's up to each of us :)

[@y2ukr87]

Well, cameras and telescopes in general are not perfect and can introduce different kinds of noise and imperfections by themselves. So, unless it is used in scientific research I do not see any harm. I wonder if someone has analyzed how many artifacts does JWST mirror introduces to pictures (I mean the hexagonal spikes caused by mirror)? As you've said, it's more of an art.

[@dankahraman354]

It is a very big problem. You should communicate with Ivo Jager creator of Star Tools. Thank you for addressing this monstrous fakery. Is this Carl fellow a master of fakery based on AI or is he a bona fide image processor?

[@Nabby13]

Hey, Cuiv! How did you manage to image the Rosette from a mirror universe? :)

[@CuivTheLazyGeek]

Mwahaha, my telescope generates the mirror universe for me to image, it's a very powerful telescope 😁 Unfortunately it keeps the light pollution from the origin universe, 😔

[@jaimepacheco721]

Wow, just keep having fun, thanks.

[@HeavenlyBackyardAstronomy]

GREAT Discussion ...period.

[@CuivTheLazyGeek]

Thanks! I'm seeing so many opinions in the comments, I really like this :)

[@CV_CA]

Suspect when they introduce details sooner or later they will be busted as someone takes a real picture.

[@bullshitvendor]

ai to correct for coma would be so awesome! i could ditch 1kg of aluminium and glass from the imaging train 😮

[@xrl7320]

Soon we'd just end up with a hollow tube and still get photos from the other universe.

[@warrenp7861]

Thought about it. I create the images because I like them and created them myself with my own equipment. Reminds me of the first time I saw the Ring nebula through my own scope. Or the planets. So...in my opinion, if you aren't creating them for scientific measurement and decision purposes...have fun and create them so you like them. My images are not intended to be scientific...just pictures to share with others the wonders of the night sky. If you decide to do imaging for scientific purposes, then do so with the utmost care that "suits the purpose" of the scientific investigations. Otherwise, enjoy!

[@CuivTheLazyGeek]

In the end it's up to each of us to do what they like :)

[@VRzichtbaar]

As an amature I aspire to get a beautifull result. To me I don't have any pretention to create a scientific result from which any astronomy conclusions can be made. By the way people over appreciate sharpness where nebula by natue are rarely scharp. Personally I focus at the denoise function only and preserve as much detail as possible. A nice topic where no one eventually will agree on. LOL. And by the way we are shifting color palets anyhow toward the visible spectrum.

[@RumourHasitYT]

I saw that image of the cigar galaxy when it was first posted. I have to say my first thought was it was fake and not appealing to my taste. However it’s all in the eye of the beholder as they say

[@leonidtalas696]

Thank you for this video. Something to think about. Lets do our best not to introduce extreme AI denoising and oversharpening to the 'scientific' part of our hobby. However, my knowledge on the following is limited, but afaik images taken by Hubble Space Telescope are processed using the Drizzle algorithm, which to my knowledge upsamples the image with AI??? Should this be considered 'fake' as well??? Or the algorithms are well known and studied, and true scientific data can still be accurately extracted from the processed image??? True, I believe we should mention what processes were used to process the image. Imagine some time in to the future, when a lot of AI processed pictures of deep sky targets will become available on the net. This large data could potentially be used to study the evolution of the details in the targets, however with AI processing etc this will become impossible to do accurately.

[@CuivTheLazyGeek]

Dithering/Drizzling is a proper mathematical operation no AI involved! It takes advantage of the information gained by fractional pixels across multiple dithered frames :)

[@beatsntoons]

I'm in the camp that as soon as you stretch your data, it's fake anyway. You're showing a dynamic range from the background that isn't really true. Most of the background extraction tools require you to remove what you THINK is background, but could actually be nebulosity. Plus how do we know how bright an object really is without working back from the inverse square law? There's so much in an astro photo that's manipulated by even people who are after "accuracy" when really everything is being change from the moment you stretch the histogram.

[@EricMilewski]

Hmm... I agree with Noam Chomsky on the 'AI' issue Generally interesting technology but way overhyped, because it really is stupid (as in not conscious) the algorithm doesn't understand WHAT! it is doing

[@peterv8]

I have it, and tried it, but I always can get a better result with the sharpening tool from PS.

[@CuivTheLazyGeek]

Good :-)

[@jreichle59]

Thanks, I use Topaz and now I know what to look out for.

[@CuivTheLazyGeek]

Thanks Joe!! That was one of the intents of the video - I underestimated how much my video could be seen as attacking people :-) It's very interesting to really look at what Topaz is doing when doing it :-)

[@philleng480]

I think the colour module in PI, that plate solves your image and then renders the stars the 'correct colour' is creating data that isnt there (photometric calibration I think?). You could do a mono image and it would fill in the colours. This crosses a line for me. At this point it might as well reference the relevant Hubble image and fill in the detail. If we want to make discoveries like the O3 nebula near Andromeda we have to believe in the data we have, not what we think we have imaged. That's my take on it.

[@davidkennedy3050]

Your take, like a lot of others is simply wrong and based on misinformation. The current Spectrophotometric Color Calibration routine in PI does not add color. You first need to convert to RGB and when you apply the adjustment gets you nowhere. The image is still mono. The now obsolete PCC does same thing.

[@gregb5149]

Yes I think its a problem when the processing goes too far, usually because the result starts to become truly ugly. I don't know what people are thinking when they process the Orion Nebula for example, with Pixinsight to such an extreme that it begins to look like its inside out and rotting! I agree that the Cigar Nebula you show looks nice but I definitely don't want to see invented detail. I think that star removal is more like trying to filter the image so that we see the subject without the foreground "noise" - its like when you take a picture of a nice scene and then discover there were telephone poles and wires obstructing the subject. You'd like to see the target without the obstructions.

[@cryptonite8495]

So Topaz is to some degree an AI art generator? I guess this was inevitable, if not Topaz than something else. As you noted, AI astro tools have been walking on the edge of that fine line, if not crossing it, for some time now. Some purists believe even stacking is fakery.

[@kekkoukedarake110]

I purchased Topaz AI Oct. 2021. I had few Wow! moments initially then start to wonder if results really true. After watching this KZbin, I realize I no longer use Topaz. That must be my conclusion without thinking.

[@CuivTheLazyGeek]

I was exactly the same as you - I personally have decided not to use TopazDenoise anymore. In the end it's all about what each of us wants to achieve with their image!

[@southbronxny5727]

It doesn't matter because the pictures are not scientific. But if NASA did excessive processing, it would be wrong. Also, the AI might be bringing out detailed information that's present in the image but that your eyes can't see because the AI is enhancing the information, not adding. Also, it doesn't matter what the star removal program adds behind the star because when the stars are added back, you won't see what is behind it anyway.

[@deepspaceexplorer4265]

We shouldn't... laziness is one thing but FRAUD is quite another. I wont touch it with a ten foot pole....

[@CuivTheLazyGeek]

I wouldn't go as far as calling in fraud!!!! Just something to be aware of? :)

[@Mr77pro]

We are taking ourselves too seriously if amateur astrophotographers start worrying about AI enhancing faint details incorrectly. We are not doing science. Repeat after me.. we are not doing science. It may seem like science because of its complexity, but its not. The Hubble Pallet is adding color that isn't actually there but nobody seems too concerned. If a real scientist uses AI sharpening on an image and claims to make a discovery that isn't actually there....then maybe its an issue. But thats not going to happen. This is art. Nobody cares if there is AI enhanced detail in our images, and if they do they should find another hobby or go get a degree. ;)

[@CuivTheLazyGeek]

Well I think it's a bit more complex than that - it really depends on what each person wants out of the hobby, and what their objectives are. And this is a hobby where amateurs can make discoveries, such as new planetary nebulae - which most definitely is science :-) Sometimes I see images so over processed with AI and other processes that they look more like paintings than photos get passed around in "normal" people Facebook who take them at face value and I wonder if that's ok. It's a difficult topic for sure!

[@Mr77pro]

@@CuivTheLazyGeek Some good points for sure! Bottom line is that nobody has a right to tell anyone else how to process their images. I stick by my anaylsis. You'll get some good traction on this video for sure😉

[@LyngJohn205]

We may be over-sensitive now because we always have to double-check and verify photos in social media to confirm that they are not faked.

[@HubertHuijbregts]

This, for me, is a question that has as many answers as there are astrophotographers. When an astrophotographer takes a photograph I consider the photograph to be her/his interpretation of the object. If details are added to make the photo more spectacular I feel that is her/his prerogative as long as she/he doesn't add details and claim a discovery of something that has not yet been seen before. What I mean to say is that art is personal. Tom might like something, Dick might hate it and Harry might think 'Mweah, I'd rather drive up a mountain and go fly my paraglider'. I am, of course, only talking about photographs that will be shared on social media, with friends or family. I want to think that I would like my photographs to be as accurate as possible but I'm not sure yet. Come to think of it now, that is not really true. Back in the day, when I used to take a lot of 'regular' photos, when I saw an image that I wanted to hang on my wall I would process it in the dark room (and later in Photoshop) to get it as 'perfect' for me as I could. When an unwanted object was in the picture that would hug all the attention I would make it disapear. I would sometimes blur the background to emphasize the object in the foreground.

[@CuivTheLazyGeek]

Why, why this reference to paragliding?? I wasn't able to fly this weekend and I'm currently having withdrawal symptoms :p Also great point about normal photography. Do we try to represent what we're seeing... Or what we're feeling?

[@HubertHuijbregts]

@@CuivTheLazyGeek So, so sorry. Hope you can take to the skies real soon.

[@in2driving]

It looks like you processed JPG files in Topaz DeNoise; these files start out being sub-optimal. Can you repeat the test with 16-bit TIFFs?

[@CuivTheLazyGeek]

Oh I did this many times on 16 bit tiffs from PixInsight, seeing exactly the same thing - when making this video I just used my latest image, whose processed version I had foolishly saved only as JPEG.

[@GrundleStiltSkin]

In my opinion, just list the processing, perhaps some terminology in a description.. false color ( FC ) Composite ( C) Artificial Intelligence ( AI) hubble pallet ( HP) ect.

[@CuivTheLazyGeek]

That's another way!

[@MaraD98]

Photography is and always has been a form of art subject to the preferences of both the photographer and prior to the digital age, the developer of the film. The image has always been dependent on the photographer's choice of lens, film / image sensor, ISO, shutter speed, perspective, time of day, angle of light, filters, etc. The developer of the film images had control over paper, exposure, processing time, sharpness, contrast, etc., whereas in the digital world those same discretions are simply applied thru software. Would we have appreciated more or less any of Ansel Adams' images if he had chosen a slightly different time of day to take each of his photos with the changes in highlights and shadows that come with each passing minute? What would we have thought if he had chosen different vantage points to capture his images or used lenses with different fields of view? For astrophotography, the end product is really not much different than that of traditional photography. The photographer gets to chose the exposure time, the framing, the colors he wishes to capture or exclude. His choice of lens, mount, camera, and shooting location all change what the film or sensor captures. The developer of those images gets to process the images in ways that change the contrast, exposure, hues, colors, saturation, etc. When light is captured outside of the visible spectrum, his or her choice of how to map the IR to visible colors is purely a form of artistic expression. In the end, there is no such thing as a "true" astrophotography image, there are just images that the majority agrees are a bit truer than the rest. And there are the images that the artists and the community find beautiful, revealing, breathtaking, satisfying and amazing. When you stop processing the image, print it or share it across the Internet, it is just a work of art that should be appreciated as simply art. While art can reveal truths, art is never purely true. Don't be hard on yourself over the art you or anyone else creates. Recognize astrophotography as art. Print it. Mount it. Share it if it is digital. And most importantly, find happiness and pride in the beauty you share with the world.

[@CuivTheLazyGeek]

I love this viewpoint, especially the relative trueness perception!

[@DB-MH11]

I always suspected this kind of fakery with AI image processing tools.

[@Ardjekwartje]

Well. People remove stars as well. To make the image more beautiful. And less sciency. 😃. So I guess everyone follows the approach they prefer.

[@CuivTheLazyGeek]

Yes, exactly

[@victorvillenapenas4274]

I personally think that the main attraction in this hobbie is being able to see and photograph the wonders of the universe, something so distant, beautiful and sometimes hard to comprehend. For me that means that the images need to be 100% scientifically accurate, whats the point in astrophotography if you are inventing detail? that image is no longer true to reality and becomes science fiction. I´ve seen 72mm ed scopes reveling details of those 1m scopes after using blurX, thats no longer YOUR image because you have basically borrowed detail from a bigger scope.

[@CuivTheLazyGeek]

I'd love to see examples from blurX that revealed details of 1m scopes! In my tests on my own images that definitely doesn't happen :-)

[@victorvillenapenas4274]

@@CuivTheLazyGeek I’ve seen a side by side comparison of a mak 150 vs a cdk 600 on the Orion Nebula and the mak was pulling detail that the cdk didn’t. I rally hope I can find the comparison again to share it with you, without any doubt BlurX does a terrific work improving the overall quality of the image. The problem (for me) is that it’s been trained on high definition images taken by the Hubble so the program already knows what’s there, traditional deconvolution may introduce fake detail but it’s working on YOUR data, it does ent know how it’s supposed to look like unlike BlurX

[@kennetknudsen7042]

another thing in AP is the extreme pixel peeping.

[@CuivTheLazyGeek]

Ha, true!!

[@tommyozzy317]

As an astrophotographer, I have some hard rules about things like this. For example, any copy and paste type stuff is a big no-no. There was a case like this years ago when someone zi think was featured as pic of the day for their milky way. Turned out that a portion of the picture was copy and pasted from a different part of the same pic. Messing with the colors, removing stars but then re-adding later are all perfectly fine for me. Of course you need to note that or divulge when being asked about it. Now doing scientific work, you need to be much more strict with any editing. Like the new O3 cloud discovered in Andromeda. Would be so easy for any of us to ramp up the processing then think we discovered some new object in the sky.

[@CuivTheLazyGeek]

I think it's great to have such personal rules - and to stick to them!

[@scottbadger4107]

I think we should stop talking about 'science'. The vast majority of astrophotographers are not using their images for any sort of scientific purpose, and even when they are, the processing is done in a way that supports some very specific scientific study, not 'for science'. The concern in our community about AI processing software and manufactured detail is really about curiosity, I think. In addition to simply taking/creating beautiful pictures, we're also curious about the astro objects we image, and because we can only 'see' these objects through our images (unlike the subjects of landscape, wildlife, portrait etc. photography), it's not surprising that there's a concern over detail that may or may not be real. The dilemma is that the more we can deconvolve/sharpen our images and bring out fine detail, the better we satisfy our curiosity about what these objects really look like, but at the same time, the more likely it is that we stray from reality..... That's the more virtuous aspect of the debate, the fat underbelly is the pervasive competitiveness in the AP community. That I can achieve with software that costs a hundred bucks what others get by spending a hundred times as much on equipment, or relocating their residence, or setting up a remote scope in some far away location, or simply spending a lot more time and effort than I'm willing to give the hobby, is seen as an unfair advantage. In that light, I suppose it is, but it's a race I'm not running.

[@CuivTheLazyGeek]

Thanks for this Scott! I think one of the fascinations I have with my own images is the challenge of getting them from Tokyo, and see structures emerge as I process the image. When I realized what Topaz was doing to the images, that didn't vibe with what I wanted to achieve - I felt like I was taking a paintbrush to the image! The only difference being that the paintbrush was AI-driven But that's for me and my images :-)

[@scottbadger4107]

@@CuivTheLazyGeek I've played with Topaz a bit, but not had a lot of success with it and assumed it was because of the type of images it was trained on i.e. not astro images. In any case, with BX the effect can be quite extraordinary, probably because I have pollution free dark skies, but terrible seeing.....so good SNR to work with and a ton of potential resolution improvement. At times, even though I can see something of an underlying structure in the original image, the sharpening is so dramatic that I check it against HST imagery, if possible, and/or dial back the settings some, but to be honest, there's also a bit of leap of faith involved..... On an unrelated note, I just happened to stumble on to your Japanese Alps video and I'm a PG pilot too! Been flying since the very start back in the late 80's..... Anyhow, the only AP connection is that I broke my back in a bad landing a little over a year ago, and of course on the same day that my brand new CEM70 arrived (and a clear sky that night to boot!), which made the recovery time seem a whole lot longer!....

[@SKYST0RY]

Unfortunately, the way the AI works can add detail and take away from the scientific value of the images. But if you are only seeking beauty, I suppose it's good enough.

[@CuivTheLazyGeek]

Very true...

[@Sidecontrol1234]

3:15 for this very reason I don't use Topaz, never been a fan of the software. Not a fan of the Cigar edit.

[@Kata_Mae]

It's like colors... Hydrogen is not red, it's pink... on low light it's brown, all LIES ! :D Actually, even what we see is arguably not reality : Some things can be seen by some and not by others, not even counting on our brain's amazing ability to make up things from pieces... wait... could it be that it's exactly what an AI does ? he he he :o

[@CuivTheLazyGeek]

That's a very good point that different people looking at the same thing will see something different! And of course we have people who can see more (or fewer) colors than the norm...

[@robertgrenader858]

All astrophotography post-processing is totally subjective. How do you know those details in Rosette do not exist? As a Landscape photographer, I can look over a scene and "read" the light. Can't do that with deep space objects. This is all a tempest in a teapot.

[@CuivTheLazyGeek]

Completely agree that post processing is subjective! For the objects I've tested Topaz Laps on, I compared with the Hubble and Hubble Palette in general - if I captured details from Tokyo that aren't visible in the Hubble images, I know something isn't quite right :D I've also seen what I know is noise suddenly become detail (but then there's always the converse, from light polluted areas, very faint nebulosity can appear as noise and be nuked during the processing). In the end we can all do what we want with our images :) I don't know whether it is a tempest in a teapot or a teapot in a tempest (those darn flying teapots hurt!) but I think it's interesting nonetheless!

[@beatsntoons]

I'm at the point where I don't care and will leave it up to the person who shares an image to decide whether to tell us whether an AI tool was used. If your goal is to show a true representation of what you captured, then it makes sense to divulge if you used an AI tool. Anything else, oh well. The cat is out of the bag now, so i don't care if anyone uses them.

[@olly7248]

No one has a right to declare that how I process my images is unethical, there’s enough policing going on in the world so please just leave people to produce their art forms however they wish… it’s not NASA or JPL doing this, it’s us 👍🏻

[@CuivTheLazyGeek]

Good point! In the end we can all do what we want!

[@astrophotonics9470]

In the end, it doesn't matter, just enjoy the Hobby.

[@CuivTheLazyGeek]

We're all here to have fun and share a sense of wonder after all :-)

[@jeffdutton1910]

if you're trying to create a "documentary" image which is to say, an accurate representation of nature then not only must you not introduce any artifacts, but you need to be confident that anything you remove from your dataset is in fact, not a part of the actual data. On the other hand, the "pretty pictures" that most of us create (and indeed, the operative word is "create") have more the nature of a caricature than a scientific measurement, accentuating features that strike the artist as important or interesting.

[@jimfitz]

People think it creates non-existent detail but I have yet to see any evidence to support that. For me, unless such evidence emerges, it brings out detail that is there but was not previously discernible. Stretching an image creates a degree of contrast that is not there; SHO creates colour that is not there; Deconvolution is an approximation that cannot accurately resurrect lost detail because it cannot accurately replicate the functions that caused the detail to be lost in the first place; … I could go on. Image processing creates a fake image, so any ‘purist’ who objects to AI is a hypocrite if they show off anything other than unstretched images. How interesting would that be?

[@CuivTheLazyGeek]

When I see Topaz creating details in images that I captured from Tokyo under terrible seeing and those details don't appear in Hubble images of the same object (including when processed from raw!), I think that's a pretty good indicator that the details don't actually exist! In the end it's up to each to decide where the line is for them! Image processing might create a "fake" image, but there is also the intent of the processing. I absolutely don't think that makes people who object to TopazDenoise hypocrites either - and I've seen multiple people, including myself, who originally didn't realize that Topaz introduced such details via "guesses" (or some might say an AI driven paintbrush), and stopped using it once realizing that. That doesn't mean other people shouldn't keep using it! As I mention in the video, no conclusion there :-)

[@jimfitz]

@@CuivTheLazyGeek maybe ‘hypocrite’ was too strong a word and I withdraw it, but I think the point still stands. People will happily accept the introduction of non-existent ‘information’ into their images (colour; contrast; sharpness; …) so there’s nothing wrong with another tool that may (or may not) do the same thing. As amateur astrophotographers we are not scientists but are more like artists - we create pleasing images, we generally don’t record anything with scientific accuracy. What constitutes a pleasing image is subjective, which is why you’ll never see two identical images even though we all photograph identical objects! Ultimately the choice of tools is also subjective, so we shouldn’t stress about it or be critical of anyone who uses a tool we don’t like.

[@woody5109]

Astrophotography has gone from a science to an art, the general public has no idea and the art sector are all about selling their product.

[@dankahraman354]

So now we are going to give up mental challenges and further develop our laziness by relying on AI???

[@astromanicdave]

You use pixinsight and call out Topaz. Ea hello!

[@CuivTheLazyGeek]

What?

[@xrl7320]

Generate art?