Photo Threads: Using AI OK?

[C0ncreteBl0nde]

Subject: POTD and AI

I wasn't sure where to post this, but it pertains to this thread specifically and I wanted to be sure that the users of this group are the ones to reply and make the decisions if any are to be made.

I'm not sure when the original rules for Photo of the Day were conceived, but I'm guessing it's been a while and they just get copied from month to month without any changes. Well, times have changed and photographers are now facing the age of AI as has recently come to my attention in the latest beta version of Photoshop and on my photo-sharing platforms.

AI is an amazing creation; I have been playing with it some and am constantly amazed at what it can do. Being a long-time Photoshop user I'm reminded of the days of painstakingly using the selection and clone tools to remove things in photos that can now be done with a press of a button. AI can be used in many ways, but what I'm referencing here with relation to this group are using AI to add/remove something from an actual photo that you took with a camera versus using a blank document in PS, typing some words in a box, hitting generate, and having Photoshop create a whole new "picture" from what you requested. There is no photography in this method at all.

I'm relatively new here (less than a year) so I'm not trying to insert myself to change the rules here, but simply asking everyone who uses and enjoys this thread: is it time to modify the rules regarding AI? What, if any of it, is okay or not okay? Is it okay to use it to remove a distracting object from a photograph but not okay to add a dog IN your photograph or post "photos" you generated by typing words in a box? What do y'all think?

I'm not sure who, if anyone still here, is in charge of this thread, but I just wanted to put this out here to garner opinions and see if any of you are in agreement that it may be time to add to the rules regarding this subject.

Thank you for letting me highjack this thread for this subject.

CB

 

[sparksd]

Subject: POTD and AI

I wasn't sure where to post this, but it pertains to this thread specifically and I wanted to be sure that the users of this group are the ones to reply and make the decisions if any are to be made.

I'm not sure when the original rules for Photo of the Day were conceived, but I'm guessing it's been a while and they just get copied from month to month without any changes. Well, times have changed and photographers are now facing the age of AI as has recently come to my attention in the latest beta version of Photoshop and on my photo-sharing platforms.

AI is an amazing creation; I have been playing with it some and am constantly amazed at what it can do. Being a long-time Photoshop user I'm reminded of the days of painstakingly using the selection and clone tools to remove things in photos that can now be done with a press of a button. AI can be used in many ways, but what I'm referencing here with relation to this group are using AI to add/remove something from an actual photo that you took with a camera versus using a blank document in PS, typing some words in a box, hitting generate, and having Photoshop create a whole new "picture" from what you requested. There is no photography in this method at all.

I'm relatively new year here (less than a year) so I'm not trying to insert myself to change the rules here, but simply asking everyone who uses and enjoys this thread: is it time to modify the rules regarding AI? What, if any of it, is okay or not okay? Is it okay to use it to remove a distracting object from a photograph but not okay to add a dog IN your photograph or post "photos" you generated by typing words in a box? What do y'all think?

I'm not sure who, if anyone still here, is in charge of this thread, but I just wanted to put this out here to garner opinions and see if any of you are in agreement that it may be time to add to the rules regarding this subject.

Thank you for letting me highjack this thread for this subject.

CB

Excellent question. We have long had the capability to modify our photos in the lab (just ask Ansel Adams) or on the computer. I am personally not uncomfortable in modifications involving removing distracting objects - up to a point. I am personally not comfortable with adding content to a photo and have never done so (building a collage is something different). A photo image that is computer generated should be identified as such - it is not "real". But great subject for discussion.

 

[Chuckeee]

Under some circumstances I think some AI functions would be acceptable for use in photo contest here.

If you are using AI to perform a function you would/could have done without AI, I would consider to be acceptable. Basically if you are using AI just as a time/labor saving tool, then it is just a tool. You the artist are still responsible for the creative process.

 

[C0ncreteBl0nde]

Thank you for moving this for me. I had company and am just now getting back to the forum.

 

[C0ncreteBl0nde]

I have been doing some thinking about this (obviously) and here's my take on this AI stuff:

We (the repetitive users of the POTD and Weekly Photo Contest threads as a whole) could decide that to use AI to remove distractions or whatever we creatively chose to remove from our PHOTOS, the same way we would have done it previously, just using a different tool and leave things as they are. OR we could allow the insertion of objects into photos (using my previous example of a dog inserted into a PHOTO to add interest) as long as it's marked as perhaps "virtual photography" and let the viewers draw their own conclusions. OR we could go as far as allowing completely artificial intelligence images as long as they are labeled as such, which would probably require a change of the name of the thread to "Image of the Day" or something broader than "Photo of the Day" since generated images using AI in Photoshop has nothing to do with taking a photo, but can be a pleasant viewing experience.

I don't know the answer; I'm hoping someone in charge or the collective rest of you, do. I'm just throwing this out there after spending a few days playing around with AI.

Please feel free to post your two cents on the matter if you have the inclination to do so. (Perhaps I am making more of this than needs to be made; I am trying to see that the rules are in place before such a photo appears rather than vica-versa.}

Thanks.

 

[r.harris1]

There's no one "in charge" here, per se. There's a broad diversity of opinions and wide-ranging points-of-view. We generally get to some sort of workable consensus, with or (preferably) without the help of MR admins.

There's photography (light through optics onto an imaging sensor (film, silicone, etc) ) and is a specific art form I enjoy. Then there's digital art, and this very broad category can certainly include photography, but can be something entirely made up in Dalle or Photoshop, or what-have-you. You're right that digital art can be a pleasant viewing experience if done well.

Photography has long been a medium where things are added and removed (since its inception). I suspect people here have swapped skies, added objects and so-forth (through gen-AI or otherwise). AI/ML techniques have long been part of our cameras and our post-processing software. Indeed, phone cameras literally require those techniques to have a workable image. I don't have any particular issue with such techniques, though I prefer that when it comes to things like "interest", it be part of the original capture, and I personally value that approach much more highly than if it gets added later. I'd like the photographer to let us know if the image is a radical departure from what they started with. If you add a generated dog, I'd love to know, for example .

There are of course endless rabbit-holes one can go down discussing acceptable manipulation of images, and tastes change. Those conversations tend to get very tedious, so I tend to have more of a "live and let live" view of things.

The broader category of non-camera-generated digital art probably needs its own sub-forum.

 

[mollyc]

Standard disclaimer, these are just my opinions, and I welcome other views.

In general, if you can do something with a manual tool (clone, crop, layer mask, etc) that is now available by AI and it is changing just a small percentage of the photo, but faster than using a manual tool, I'm okay with "AI." I actually haven't tried generative fill in PS, but if I needed to remove say, a stop sign, and AI did it better than using the clone tool, I don't think that should be disallowed from the POTD thread.

I'm okay with sky swaps (though often feels the original image even with a blown sky looks more natural) and just about anything that can be done with a manual tool in PS; almost all legacy PS tools are built around darkroom tools. The POTD is not a photo competition thread, but it is a photo-based thread and I believe that the image should originate as a photograph. And similar to what @r.harris1 said above, if someone wants to start with a "real" photograph and then generate in a dog, just please disclose that. I would say these guidelines should be in place on the weekly competition threads as well, although for that I could see why some members might want even stricter rules with zero generation or AI (truthfully I'd probably err on the side of more strict rules than less on the weekly threads).

 

[sparksd]

I like POTD as I hope it helps me a better photographer, not just in capturing the original image but also in post-processing. When people post photos with significant changes applied to the original, it would be interesting from a learning perspective to have that pointed out. It would not be something to be criticized for.

Digital manipulation of photos is an interesting issue that has been heavily argued. The famous outdoor photographer Art Wolfe got into some hot water over it in one of his books, Migrations:

The call of the wild beckons photographer Art Wolfe

 

[mollyc]

Personally I think if someone wants to share a heavily AI-ed image they should make a new post. Or have an AI sharing thread. As commented earlier, it's a slippery slope with a lot of rabbit holes (how many more euphemisms can we use), and AI isn't inherently bad, but it is very different, and as such needs to be noted.

 

[cjsuk]

Interesting thread. I only use AI for denoise/selection at the moment. Anything else feels like I'm giving part of my soul away.

 

[bunnspecial]

Fundamentally I think the question is not one of AI, but one of manipulation in general

1. How much can a photo be manipulated and still re main the creator's work?

2. How much can a photo be manipulated and still be a photo?

This is a debate that's as old as photography, and as others have said is one that has been around forever. Compare a contact print(if you can find one-I've seen ONE) of Moonrise, Hernandez to the actual print. For that matter compare say a print made in the 50s to the 80s as Adam's interpretation and printing preference for that negative changed over time, but he had to do a LOT of work to get everything to print.

I've know people to argue that even adjusting exposure is a form manipulation. Some of the people who I've heard argue this were film purists...who only shot print film that was printed at a lab(and yes, they're adjusting your exposure). The only people with a moral high-ground so to speak on exposure adjustments are the ones who shoot/shoot slide film or who use JPEGs straight out of the camera. Others consider pretty much anything fair game, while I'd venture to say most of us fall somewhere in-between.

If you're using AI tools to do the same kinds of manipulations you did without them, I say more power to you. If we wanted to ban AI in POTD, we'd have to ban all manipulation IMO. If something gets heavily changed to the point where it's effectively a new work, well that could be a different discussion.

For me, my usual manipulations are exposure, color, cropping, and sometimes perspective adjustments. The only AI tool I've used is noise reduction, which I find works better than any other noise tool I've used. I played with blur background, but found it looked a little too much like the results from phone cameras for my liking...

 

[sparksd]

Fundamentally I think the question is not one of AI, but one of manipulation in general

1. How much can a photo be manipulated and still re main the creator's work?

2. How much can a photo be manipulated and still be a photo?

This is a debate that's as old as photography, and as others have said is one that has been around forever. Compare a contact print(if you can find one-I've seen ONE) of Moonrise, Hernandez to the actual print. For that matter compare say a print made in the 50s to the 80s as Adam's interpretation and printing preference for that negative changed over time, but he had to do a LOT of work to get everything to print.

I've know people to argue that even adjusting exposure is a form manipulation. Some of the people who I've heard argue this were film purists...who only shot print film that was printed at a lab(and yes, they're adjusting your exposure). The only people with a moral high-ground so to speak on exposure adjustments are the ones who shoot/shoot slide film or who use JPEGs straight out of the camera. Others consider pretty much anything fair game, while I'd venture to say most of us fall somewhere in-between.

If you're using AI tools to do the same kinds of manipulations you did without them, I say more power to you. If we wanted to ban AI in POTD, we'd have to ban all manipulation IMO. If something gets heavily changed to the point where it's effectively a new work, well that could be a different discussion.

For me, my usual manipulations are exposure, color, cropping, and sometimes perspective adjustments. The only AI tool I've used is noise reduction, which I find works better than any other noise tool I've used. I played with blur background, but found it looked a little too much like the results from phone cameras for my liking...

Adams did do a lot of work -

https://www.haroldhallphotography.com/ansel-adams-the-making-of-40-photographs/

 

[Clix Pix]

In my opinion, "digital art" is something quite different from "photography" and usually involves much more than simply shooting and editing an image. Digital art often involves significant manipulation of the image in ways which may actually alter its meaning as well as appearance. AI certainly is used in digital art and that's fine.....as long as the resulting product is identified as such: either digital art or AI-enhanced art, whatever the artist wants to call it.

Some of us use software other than PS or LR to edit our images. I tend to be rather simplistic in my approach to editing and presenting an image, and don't get into working with layers or swapping/adding skies, objects, creatures or people into an image. I will, however, remove an object from a scene, such as the tail-end of a bird which swam or flew into the scene just as I was shooting another bird, or that trash bin that I couldn't remove physically from a scene in a situation where I also couldn't get a different, better camera angle in order to avoid the trash bin altogether. Stray tree limbs or branches, odd bits-and-pieces of paper on the ground near my subject also are fair game for quick, painless removal. I am not keen on overly-colorful HDR sorts of skies and scenes -- they look fake right from the get-go.

Once about a year or so ago just for fun I did try out one of those AI programs that was on some online site to experiment with an image just to see how it worked and what would happen. It was a fun little novelty. Haven't bothered with that again since.

A couple of the software programs I use for editing do have AI functions, and programs such as Topaz AI or Topaz Gigapixel, definitely make AI a big part of their reason for being. I use one or the other only for one thing, and that is when I need or want to increase the resolution in an image in order to have more cropping flexibility. An example of that would be the photo of the cormorant with the fish in mid-air about to land in his mouth. I upscaled by 2x so that I wouldn't lose any resolution while cropping so that the primary subject was much more visible. The scene is exactly as shot, except that I cropped in more tightly than usual in order to show off the subject in closeup.

In general, I think that if someone does present an image which he or she has significantly manipulated with AI or especially if it was wholly created by AI, then this should be stated. Significant manipulations go far beyond simple cropping/object removal, etc. and these days AI is becoming so realistic that it can be difficult to identify an AI-created image. IMHO that is NOT photography and does not belong in the POTD thread at all.

In another photography forum to which I belong, some members have been experimenting with AI-generated images and have started threads with them; in each situation the OP clearly identifies them as such. So far there haven't been too many of those images or threads there. I think some people tinker with such things a few times and then move back to doing "real" photography.

As for our POTD, it has definitely changed in character and broadened in scope in the years since it was first created, and the originator of the thread is no longer a member of MR. Hence, there is no "leader" here, no one member responsible for running things, except that occasionally when we do have issues or request a change, one of the admins or mods does step in to attend to the situation.

The POTD thread started out as one in which those who were using (the then-newly becoming popular) digital cameras would have a thread in which to display their best work, their most interesting sights and scenes from travels, some interesting scenes and sights from daily life: whatever they were currently shooting with their new gear. It was meant to be distinct from the other section on MR where people share their snapshots of their cats, dogs, cars, kids, breakfasts/lunches/dinners/snacks/beverages, etc. However, through the years there has been some overlap.

The POTD thread and the entire Photography section has evolved over time, but yes, we do need to be careful about how far we go with AI. Let's keep the POTD thread about actual photography! I agree with r.Harris1, who wrote: "The broader category of non-camera-generated digital art probably needs its own sub-forum."

 

[TechnoMonk]

personally, i post unedited pics in Potd, just the way it was captured. I am not against AI generated or edited pics, if they are identified. I use topaz AI for video and is great for upscaling.

 

[r.harris1]

Adams did do a lot of work -

https://www.haroldhallphotography.com/ansel-adams-the-making-of-40-photographs/

“Making of 40…” is a terrific book. Yes, he did a ton of work!

 

[katbel]

We had a similar discussion sometime ago- before AI- about photos for a contest published on some famous websites where most photos were elaborated, some heavily.
The answers from you guys were very similar to the ones you are writing today.
I'm still considering a difference between Photos and Art Photos.

 

[AlixSPQR]

Excellent question. We have long had the capability to modify our photos in the lab (just ask Ansel Adams) or on the computer. I am personally not uncomfortable in modifications involving removing distracting objects - up to a point. I am personally not comfortable with adding content to a photo and have never done so (building a collage is something different). A photo image that is computer generated should be identified as such - it is not "real". But great subject for discussion.

It would be hilarious if someone added a distracting object that most would remove. That would enhance the photo’s credibility. The irony.

 

[Apple fanboy]

Personally cloning out a small distracting object in a shot is fine. Swapping a subject from one scene to another or swapping a sky is too far for my tastes.
I’d rather see a separate section or thread for digital manipulation. It can have its place, but taking a bird from one image, a tree from somewhere else and a sky from somewhere else completely different and merging them is not photography. It’s something different.

 

[mollyc]

Personally cloning out a small distracting object in a shot is fine. Swapping a subject from one scene to another or swapping a sky is too far for my tastes.
I’d rather see a separate section or thread for digital manipulation. It can have its place, but taking a bird from one image, a tree from somewhere else and a sky from somewhere else completely different and merging them is not photography. It’s something different.

I respect your opinion and my thoughts don't deviate too far from them, but there is evidence of hand manipulation of photos for just about as long as photography has been going on. Even well regarded historical images can have a lot of parts swapping.

12 Historic Photographs That Were Manipulated

With the advent of powerful desktop publishing software like Adobe Photoshop, digital image manipulation has become mainstream. However, even before the invention of such software, doctored …

twistedsifter.com

 

[Apple fanboy]

I respect your opinion and my thoughts don't deviate too far from them, but there is evidence of hand manipulation of photos for just about as long as photography has been going on. Even well regarded historical images can have a lot of parts swapping.

12 Historic Photographs That Were Manipulated

With the advent of powerful desktop publishing software like Adobe Photoshop, digital image manipulation has become mainstream. However, even before the invention of such software, doctored …

twistedsifter.com

And that’s fine too. But if an image I shot bares no resemblance to the one I end up with that isn’t just photography. It’s a different art form to me.

 

[Jumpthesnark]

What a great topic for discussion. Thank you @C0ncreteBl0nde for starting this thread.

My background is in journalism so the tools I use to alter an image have always been minimal at best, and always just to "fix" an image so that it looks the way my eyes saw the moment when I shot it. So I've never cloned out part of a picture, added an element that wasn't there, etc. I stay away from heavy dodging and burning. In the POTD forum I have converted some images to b/w if that was the way I conceived of the picture when it was shot, though. I will adjust exposure, contrast and color temp so that it looks closer to what my eyes and brain saw. So the post processing I use tends to push the image away from my failures as a photographer (and limits of the gear) to get the final image to appear closer to the way I saw it.

But I have gotten into astrophotography in recent years and many of those star trail, meteor shower or Milky Way photos do require multiple stacked images, and I have learned Photoshop and LR techniques I never used before. In those cases I'm thinking of the work that I produce as artistic and not journalistic, so I'm applying different rules to those images.

For example, with a long exposure my cameras can see the Milky Way in much greater brightness and detail than I can with my naked eye. In this case, it's an opposite approach from the rest of my work. I'm pushing the gear and using the software to create something that I could not see myself. Though those things (the Milky Way, meteors, stars) existed right in front of me.

That's a long way of saying that I have never used AI tools in post for most of my work and don't intend to. Will I use an AI tool if it makes an astrophoto better, or it makes astro post easier? That's much more likely.

Usually in POTD I'll add information to the image such as the camera body, lens and settings I used. If I do decide to use any AI in my astrophotography, I'll mention that in the caption as well. That may be all the guidance that is needed in the forum - for people to mention their use of AI in the images they post.

I agree with @Clix Pix who agreed with @r.harris1 - entirely AI-generated images need their own sub-forum. I don't think of those images as photographs at all, but that shouldn't stop people from creating and sharing them here.

 

[_timo_redux_]

I like for the intentionality of the photograph, when it was taken, to read in the result. So my rules, which are personal, mean no sky swapping or "extra" cloning, and no AI-creation. I actually prefer some imperfect elements to remain in the shot: as usually at the time of capture I try to frame my picture to exclude them, but if I can't, it's a part of the photo. Even in cropping, I have a personal rule to not e.g., change landscape for portrait, though either to square is OK.

I'm also of the opinion that trying to perfectly erase problems with a photo, including lots of AI-assistance, changes the gestalt enough such that the result often feels "off."

But of course, as with all rules, some can and should be broken, always in service of a better* picture.


*there's a lot to unpack in what makes a picture "better"

 

[katbel]

I respect your opinion and my thoughts don't deviate too far from them, but there is evidence of hand manipulation of photos for just about as long as photography has been going on. Even well regarded historical images can have a lot of parts swapping.

12 Historic Photographs That Were Manipulated

With the advent of powerful desktop publishing software like Adobe Photoshop, digital image manipulation has become mainstream. However, even before the invention of such software, doctored …

twistedsifter.com

Very interesting, thanks for the link!

 

[bunnspecial]

I think this thread just reinforces my earlier point that we all have a line about what's okay or not to change in a photo, and I don't think AI changes that.

I'm not someone who is big on removing items unless I can remove them by cropping. Even cropping-to me-is questionable about how much you consider it "manipulation" especially if you intend to print since the aspect ratio of most sensors(and 35mm film) doesn't match most of the common print sizes. If you have a 35mm negative, or a 6x6 negative, and want to print an 8x10, you have to crop something, it's just a question of what. The same is true of someone using a 16.9x24mm or 24x36mm sensor.

BTW, I got adventurous once and decided I was going to try and learn hand retouching. Packed away somewhere, I have a retouching table-a nifty little 8x10 lightbox that vibrates gently to help you actually put your retouching marks on the film with a soft lead pencil. I have a box of 4x5 Kodak TXP320, along with probably a dozen 220 rolls of the same-this is the only film I'm aware of that's made with a "toothed" texture on the base for hand retouching. I have a bottle of treatment to apply that to any film, although obviously it's easiest to start with film that's already that way. My attempts at doing so were an abject failure to put it mildly.

 

[splifingate]

I have been doing some thinking about this (obviously) and here's my take on this AI stuff:

We (the repetitive users of the POTD and Weekly Photo Contest threads as a whole) could decide that to use AI to remove distractions or whatever we creatively chose to remove from our PHOTOS, the same way we would have done it previously, just using a different tool and leave things as they are. OR we could allow the insertion of objects into photos (using my previous example of a dog inserted into a PHOTO to add interest) as long as it's marked as perhaps "virtual photography" and let the viewers draw their own conclusions. OR we could go as far as allowing completely artificial intelligence images as long as they are labeled as such, which would probably require a change of the name of the thread to "Image of the Day" or something broader than "Photo of the Day" since generated images using AI in Photoshop has nothing to do with taking a photo, but can be a pleasant viewing experience.

I don't know the answer; I'm hoping someone in charge or the collective rest of you, do. I'm just throwing this out there after spending a few days playing around with AI.

Please feel free to post your two cents on the matter if you have the inclination to do so. (Perhaps I am making more of this than needs to be made; I am trying to see that the rules are in place before such a photo appears rather than vica-versa.}

Thanks.

We [TINW]; we AWTP!

Because: USENET....

*cough*, I digress

This bandying-about of the acronymical "AI" thing is rather tough, as the thingies we have at our disposal, today, are not--in any way--intelligent; they are merely human-made, computational constructs.

When Nigel Sheldon and Ozzie Hernandez Isaacs cyphered the physics:hardware that could compress time&space--such that instantaneous wormholes between one time:space could be directly connected to another--this "leap" was only able to happen because of the computational capacity that existed during this time (c.AD2050, IIRC).

Out of these sophisticated arrays developedevolved silicon-derived intelligence (SI). and--through negotiation with humans--RI (Restricted Intelligence), upon which hence-forth became the formation of "AI"-level soft, was established.

But, I progress

On OT:

Given advanced modelling/modification/editing, I don't know how we--as consumers--can really tell.

Seems like a waste of time to spend so much time falsing/modding things 🤷‍♂️

It's far easier for me to re-edit my submission--admitting that I am really bad at sexing insects--than it would be for me to alter the photo of a moth I submitted to widen the antennae.

For that matter, why would I share a photo of a moth that I created with some OpenArse pay-per-view app with no empirical context?

Did I mention that I'm really, really REALLY bad at sexing sub-/species