I am not a fan of AI in photography, but I do like to keeping up with technology. Yesterday I was playing with a new image editor from google. The article I read called this app a photoshop killer. I was curious and thought I would give it a try I uploaded an image and typed in "wrap the dog in a blanket" Way too easy. I know I will not be jumping on board with this app. I am curious when this Image falls. Is it AI, or just retouched. I could of probably done the same in photoshop in a few days, and it would be a lot more fun.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Use of AI in the MR photography forum
- Thread starter mollyc
- Start date
- Sort by reaction score
Personally I would categorize this as AI generation, not just retouching.
It’s an interesting question that @dimme raises. If you can do it in non-AI photoshop (photoshop can be a verb for a reason 
), is that acceptable? If I spend a few days adding a sweater or blanket on an image I photographed of my dog, cat, gerbil, pet goat, etc, where does that fit in? No AI used, just photoshop and its many non-AI tricks.
), is that acceptable? If I spend a few days adding a sweater or blanket on an image I photographed of my dog, cat, gerbil, pet goat, etc, where does that fit in? No AI used, just photoshop and its many non-AI tricks.
For me adding to a photo changes it more than taking away. The image on the left is representative of what you saw. The image on the right is representative of an idea you had that you added in later (regardless of AI or PS being used).It’s an interesting question that @dimme raises. If you can do it in non-AI photoshop (photoshop can be a verb for a reason
Subtraction from a photo is different. If it’s impossible to get the view you want without strangers in the shot or that power line cutting across the scene, that’s a bit different (for me anyway).
Clever app though. I do find it funny when people advertise their product as a killer.
PS killer
iPhone killer.
Tesla killed.
Mostly they just aren’t. The feature to add a blanket (or whatever) to an image certainly won’t get me cancelling my subscription. But I’m sure it will appeal to some.
I agree.
I would say not acceptable. By adding the blanket the image has changed character and becomes something completely different, whether by AI or Photoshop skills…It’s an interesting question that @dimme raises. If you can do it in non-AI photoshop (photoshop can be a verb for a reason
Did the dog have a blanket in the original and you — example — changed the colour of the stripe? To me that would be OK.
100% Agree.
Anyway, I suspect a lot of this will come down to people's — can I say morality? I am quite sure that people will use AI and not declare it. I could call it cheating, others won't.
Such is life.
Last edited:
To be clear, I agree with the consensus here but I'm pushing a little because it's an interesting discussion (to me) . It seems clear then that even removal of objects would have its limits. If I remove the dog's blanket (if it had one originally), that's too far, I should think, since I have to replace the dog's body with a made up one, but what about removal of the blue leash? That's akin to removing trash, power lines, annoying people and such. Many of those artifacts could be removed with "traditional" photoshopping techniques like cloning. They could also be removed with AI.
. It seems clear then that even removal of objects would have its limits. If I remove the dog's blanket (if it had one originally), that's too far, I should think, since I have to replace the dog's body with a made up one, but what about removal of the blue leash? That's akin to removing trash, power lines, annoying people and such. Many of those artifacts could be removed with "traditional" photoshopping techniques like cloning. They could also be removed with AI.
Last edited:
If the leash was obtrusive I'd treat it the same way I do powerlines, a piece of trash on the ground and the like. I'd nuke them without any fear that I am altering the intent of the image. Whether that is by traditional methods or the software's built-in AI tools.To be clear, I agree with the consensus here but I'm pushing a little because it's an interesting discussion (to me)
************
A so-so pic of a dog on the beach altered / interfered with until it becomes "😍😍😍 Oh look at my dog on the beach in his cute blanket! 😍😍😍"
That would be underhand IMHO, no matter whether you did it manually in Photoshop or asked ChatGPT to do it.
If you want a photo of your dog in a blanket get a damn blanket, wrap it around and take the photograph. Don't fake it.
I think that one has to differentiate between A.) advanced statistical methods and models (the later being machine learning) and B.) generative AI.
A.) Stacking of Photos, deconvolution, techniques to improve local contrast, HDR, etc. - basically everything from the Image Processing Handbook is perfectly fine IMHO.
B.) replacing part of an image like the towel-wrapped-dog above moves out of the realm of photography into Art - that’s a photorealistic artistic image.
One difference is that you are not fully in control of all parameters as in A.) and you are letting model replace a large part of the object in focus changing the information carried ultimately by the photorealistic image.
You need that as an illustration? Perfectly fine, but no longer a photo (taken).
It is IMHO exactly the same like when Samsung was caught faking the moon zoom photos in the mobile phone image processing pipeline.
nota bene: anything B.) should be cleared labelled or coming with a caption. IMHO. YMMV.
A.) Stacking of Photos, deconvolution, techniques to improve local contrast, HDR, etc. - basically everything from the Image Processing Handbook is perfectly fine IMHO.
B.) replacing part of an image like the towel-wrapped-dog above moves out of the realm of photography into Art - that’s a photorealistic artistic image.
One difference is that you are not fully in control of all parameters as in A.) and you are letting model replace a large part of the object in focus changing the information carried ultimately by the photorealistic image.
You need that as an illustration? Perfectly fine, but no longer a photo (taken).
It is IMHO exactly the same like when Samsung was caught faking the moon zoom photos in the mobile phone image processing pipeline.
nota bene: anything B.) should be cleared labelled or coming with a caption. IMHO. YMMV.
Last edited:
As an Amazon Associate, MacRumors earns a commission from qualifying purchases made through links in this post.
An excellent and timely thread and an absolutely fascinating discussion, conducted, yes, with some heat but entirely without rancour; I'd almost recommend making a sticky of it, so that it can be consulted, cited, thought about, - kept where it can be seen - and added to, as needed.
Retouching a photo to remove unwanted objects from the background or foreground, transforming the sky and things like that makes no difference if done with AI or manually. That's up to the individual, and the situation. For example, a portraiture photographer: in this case his or her client may or may not want to show skin imperfections such as pimples, scars, and so on in his or her photo. Does it matter if the photographer uses an AI app to remove much of the imperfections in a second or two, or just to take care of it manually-still using an app plus spending a long time doing it manually?
Regardless, what is most important to the individual taking the photo? a. for the photographer to feel pleased with himself of herself about the photo, or b. for how the viewer (or client) "feels" about the photo?
See...all is based on emotions, something every one of us has and expresses.
Regardless, what is most important to the individual taking the photo? a. for the photographer to feel pleased with himself of herself about the photo, or b. for how the viewer (or client) "feels" about the photo?
See...all is based on emotions, something every one of us has and expresses.
In POTD rules thread, I also have proposed a change of the second rule of the POTD thread text that is copied and pasted to start the thread in a new month. The second bullet point of the guidelines has been corrected to include All photos should be your own work and not AI generated unless special circumstances have been agreed upon.
Same goes with the photo contests.
Try the Picture Gallery forum:
As I post this, I see threads for posting pictures of: Mac setups, pets, desktops, watches (trad only), last purchases, meals, beer, cars, fountain pens, and more. That's about as diverse a set of subjects as you could find, and I see no obvious problem if a thread for AI-generated pics is added.
This is just one moderator's personal suggestion. It may be overruled, but it's worth a shot.
i don’t know but if you can’t find one you should start it.