Should AI edited photos be labelled?

[vkdev]

Advertisement (gone after registration)

I agree.
As long as it is creativity, any means are acceptable, and it doesn't need to be labeled. Does it really matter to you what the actual color of the dress is on the woman in the photo, if the new color makes the photo more beautiful?
When it becomes an attempt to manipulate people (for example, a president with his mistress, or piles of bodies in the street, or false advertising of a hotel or a tourist trip) — that's a crime, and the crime is not assessed by the presence of a visible watermark.
btw if the photo is real, you can present the original raw file with a "Leica Content Credentials" (etc).

[Advertisement]

The Leica Q3 is available for pre-order at B&H Photo Video and Adorama

[jaapv]

6 hours ago, frame-it said:

if its done it should be mentioned in some way

If what has been done and why?  
If an image is presented as an exact replica of reality yes, all edits, even the most minor ones, should be mentioned.  If it is presented as the photographer”s interpretation and vision, there is no need, but there may be limitations like on this forum: significant involvement of (Leica gear, or in a competition: no object removal, etc.
If it is a new creation of the photographer”s reality it may be interesting to follow the photographer”s creative journey (pgb”s case) but there cannot be an obligation. 

Summarizing my personal opinion: Only if a photograph has been deliberately been modified to misrepresent reality and intended as a “lie”  there is an obligation to make the viewer aware of the fact. 
Whether the method was traditional or the latest has no relevance. 
This was the case 100 years ago, from the beginning of digital photography and today. Nothing has changed but the methods used. The results are the same and the result is we call a photograph. 
 

https://serenademagazine.art/photo-manipulation-before-photoshop-the-art-of-darkroom-myths/

[frame-it]

32 minutes ago, jaapv said:

If it is presented as the photographer”s interpretation and vision, there is no need,

 

32 minutes ago, jaapv said:

Only if a photograph has been deliberately been modified to misrepresent reality and intended as a “lie”  there is an obligation to make the viewer aware of the fact. 

agree

[01maciel]

There seems to be a need for discussion on this topic apparently. And it's even quite controversial which is good. The biggest difficulty I see is that commercial image editing software is now being infiltrated by AI. I deliberately use the negatively connotated word “infiltrated” because users no longer know whether AI functions are being used when they edit an image. This starts with noise reduction, for example. Samsung apparently has a function that automatically replaces frequently photographed subjects that have been photographed unfavourably with a better moon (e.g.) fished out of the net by AI. If you don't know that, how can you be honest with forum users and label your image accordingly? So you no longer really know what the software is doing in the background. And basically, you are no longer the master of your own image. My personal opinion is that software manufacturers are leading us users by the nose.

It is said that AI in photography is possibly being more accepted by the younger generation of photographer. In turn it means Leica users are immune to this:) But I don't think this is the whole truth.
But on the other hand, the photos taken by conventional amateur photographers using AI are getting better and better and raising the bar higher and higher. In the end, those who don't use AI are at a disadvantage. In the hobby sector, they don't get any likes (so what but it could be frustating;), and in the commercial sector, they don't get any customers. They prefer to use the colourful and perfect images available on iStock.

Edited September 17 by 01maciel

[hansvons]

5 hours ago, jaapv said:

With all due respect, this is whataboutism. 

@pgh No it‘s not. Authoritarianism is only possible if the electorate is fine with tossing away universal values such a facts, truth, science, diversity, openness, you name it.

This is precisely what‘s happening in front of our eyes in the US right now. My personal assessment: museums, galleries, theatres, the film industry, anything dealing with culture of any sort will be next in line to feel Trump‘s sledge hammer. And there will be hardly any resistance because  the majority of the American people have already dumped their values or never had them in the first place because they are undereducated.

Similar developments are happening in the UK, Germany, France, the Western EU in general and, already more advanced, in Central Eastern Europe.

Now AI. It‘s the great divider, truth destroyer, leveller because as god it knows everything including you and tells you only what you want to hear—no universal truth required only a servile friendliness covering unknown agendas. And, for the few AI-literate, it‘s the path to even more performance, riches, distinction to the rest. It can be brilliant for science, medicine, engineering. Not sure if that balances out the unprecedented societal dangers.

For photography it‘s the nail into the coffin. Computational photography on my phone (eg, the Leica photo app) is mental. The results are breathtaking in an online media context, which is 99,9% where photography takes place.

Printing is a different game. But do I want to nail perfect, clean pictures onto my wall? And here it get‘s interesting for me, and that’s where AI loses out big time. Authenticity is king in my world. That‘s the one thing AI cannot do. Most of my printed images have been shot on negative, none of them "leveraged“ AI. Not because of philosophical concerns (art is free, do whatever you please to do) because I don’t like it. 

[01maciel]

16 minutes ago, hansvons said:

This is precisely what‘s happening in front of our eyes in the US right now.

That is a different matter; political discussions are out of place here. So please, don't and stop it.

 

55 minutes ago, jaapv said:

Only if a photograph has been deliberately been modified to misrepresent reality and intended as a “lie”

As is so often the case in legislation, the difficulty arises from the so-called rubber clauses. What exactly does "deliberately been modified" mean? Everyone interprets them in their own way, but some do not interpret them at all. And because this remains unclear and there will be no jury, these well-intentioned intentions remain mere theory.

[jaapv]

Advertisement (gone after registration)

AI is indeed the negative buzzword of the day. But it is in fact completely useless as an ethical touchstone as there a endless variations. In the end it is nothing more than the gathering and integration of as much data’s possible. 
The real question is: how are we using this evolving technology? Is the data gathering lawful? Noting new there, up until now we did it ourselves using Google and nobody managed to answer that question over thirty years. Next dilemmas: Is the compilation of data trustworthy? Clear answer: no in many cases without human interaction it is prone to hallucinate Next question: how do we use it? There we come to the photographic bit: It falls into two swo categories ( and an inbetween causing confusion)

1. Using the data gathered to refine editing without altering the essence of the image eg. Noise reduction, making selections, etc.

2. Using the data gathered to create new content: creating background when removing objects, adding content, sky replacement ( although we had that in a simple form before, so maybe it falls under #1) , neural filters, etc. 
 

And then there is the confusing in-between: using the image content to generate new reality eg content aware tools like cropping, other content aware tools, resolution multipliers, lens corrections, etc. Even dust removal and cloning could be mentioned  

I don’t see any harm in the use of #1 as it is no more than refining the editing as we have been doing since Photoshop was conceived; the main problem is the use of # 2. To what extent is that ethically admissible without disclosure? There cannot be a sensible answer as it is completely subjective and determined by intent. As mentioned in another post, no line can be drawn. 
As for the in-between, we have been accepting this since 1991 in the digital realm and since mid-nineteenth century in photography in general. 

So I see no case or at least possibility for regulation, other than demanding that the raw file or negative be presented or -hopefully- authentication becoming universal. 

[hansvons]

10 minutes ago, 01maciel said:

That is a different matter; political discussions are out of place here. So please, don't and stop it.

No it‘s not political. It‘s a societal/historical observation, shared by all publicly speaking, reputable historians, from the left (Timothy Snyder) to the right (Francis Fukuyama, Neil Ferguson). We are talking science here, basically. What you make of it is entirely up to you. 

[hansvons]

11 minutes ago, jaapv said:

AI is indeed the negative buzzword of the day. But it is in fact completely useless as an ethical touchstone as there are endless variations. In the end it is nothing more than the gathering and integration of as much data’s possible. 
The real question is: how are we using this evolving technology? Is the data gathering lawful? Noting new there, up until now we did it ourselves using Google and nobody managed to answer that question over thirty years. Next dilemmas: Is the compilation of data trustworthy? Clear answer: no in many cases without human interaction it is prone to hallucinate Next question: how do we use it? There we come to the photogenic bit: It falls into two swo categories ( and an inbeteeen causing confusion)

1. Using the data gathered to refine editing without altering the essence of the image eg. Noise reduction, making selections, etc.

2. Using the data gathered to create new content: creating background when removing objects, adding content, sky replacement ( although we had that in a simple form before, so maybe it falls under #1) , neural filters, etc. 
 

And then there is the confusing in-between: using the image content to generate new reality eg content aware tools like cropping, other content aware tools, resolution multipliers, lens corrections, etc. Even dust removal and cloning could be mentioned  

I don’t see any harm in the use of #1 as it is no more than refining the editing as we have been doing since Photoshop was conceived; the main problem is the use of # 2. To what extent is that ethically admissible without disclosure? There cannot be a sensible answer as it is completely subjective and determined by intent. As mentioned in another post, no line can be drawn. 
As for the in-between, we have been accepting this since 1991 in the digital realm and since mid-nineteenth century in photography in general. 

So I see no case or at least possibility for regulation, other than demanding that the raw file or negative be presented or -hopefully- authentication becoming universal. 

Agreed.

[jaapv]

47 minutes ago, 01maciel said:

As is so often the case in legislation, the difficulty arises from the so-called rubber clauses. What exactly does "deliberately been modified" mean? Everyone interprets them in their own way, but some do not interpret them at all. And because this remains unclear and there will be no jury, these well-intentioned intentions remain mere theory.

Clear cases are present, war-time propaganda and Stalin's airbrush ( see the article I linked) come to mind. It gets difficult when we enter manipulation by omission: where do you point your camera and how do you crop? 

[LUF Admin]

Some official remarks from the admin.

In this forum please leave out political discussions. They are important but this forum is not the right place. We tried this out, several times, always with the same result that political or other sensitive topics lead to conflicts. Please don’t.

AI Images are a bit more complicated.

For the Leica Forum I have these guidelines:

1. No AI generated images. As mentioned before, AI is no Leica product, so not allowed. But there are more reasons: If we allow or tolerate AI images this will lead to a general distrust if images are „real“ (= made in a photographic process). Such distrust can harm climate of discussion.
2. No problem with AI help in post production or enhancement. A lot of cameras have built in AI trained algorithms and most post processing software apps use AI  for denoising, sharpening and upscaling. Don't hesitate to use it.

Between these extremes we have a grey area:

• Removing objects via AI (or by Photoshop manual work). In my personal opinion (not as admin) OK.
• Adding objects with AI. Not OK. I had a similar discussion in one of my other discussion boards about a forest image. The photographer had added a small squirrel in the foreground. Only 0,5% of the image were affected – but the main subject was changed.

My suggestion would be not to allow such alterations.

Andreas

PS.: No politics. Not at all. Thanks.

[jaapv]

Or if you want to blend (Leica!) images - AI or non-Leica generated blends are a no-no as Andreas says- DISCLOSE!!  As we have no way to detect it, moderation cannot monitor this, so it is an honesty thing. Tricking people has no place on this forum. 

My personal opinion. 

[Dazzajl]

Photography has been smoke and mirrors since its inception and it’s no different now. As guests in a private space, it’s our duty follow the house rules, so whatever is decided is correct in this case. 
 

As to the discussion on whether a photograph presents a reality or a lie, there have been both since the very beginning. Showing several seconds, minutes, even hours in a single frame is a complete distortion of reality. Using wide or telephoto lenses to provide perspectives that are far from our natural view mislead the viewer and deliver an artistic intent. Dodging and burning both serve to remove reality from the frame. As a printer I would often remove objects from prints that are in the neg, bleaching out and spotting in are century old techniques now.

I think the idea that we can define one thing as being a photograph is also an illusion. 

[01maciel]

Honesty and transparency are the foundations of this forum. I think we can all agree on that. But perhaps, this may be undermined, not by forum participants, but by the way in which the image editing software offered are used. 
I hope that the general rules regarding AI are well known (Andreas has pointed this out again here). However, my original concern was whether photos that have been edited with AI-Tools should also be labelled as having been edited or enhanced with AI tools.
Adobe Photoshop for example offers a whole range of AI- tools that can be easily applied in the normal image editing process without the user really knowing what is going on. There are tools like enhance productivity and creativity through intelligent features like Generative Fill and Generative Expand for seamless content generation and image extension, Automated Masking, AI-based noise reduction, or Neural Filters for one-click complex edits. These AI-driven capabilities automate tasks and improve image quality. 
If these tools are used, either in whole or in part, isn't that already deceiving the subsequent viewers of the image if it is not indicated separately? 
I think it is a matter of personal judgement, to which there is probably no definitive answer. And presumably everything has already been said on this. 

Or almost  Simply saying that it has always been this way and will remain so is, in my opinion, too simplistic and ignores the reality of technological advances.

IMO the actual difference between normal image editing and the use of AI tools is a fine line, but it should be such that in the former case, the user knows what they are doing, for example, tinkering around in the RGB colour space of the image. Using AI tools means relying on the probabilities that will produce a good image, the latest and greatest can be downloaded from the internet in the background silently at any time. A user relinquishing control over the editing process in favour of a good image. I think that's the real point. Isn't there a danger that at some point you'll just become a button pusher without knowing what you're actually doing?

 

[Stephen.s1]

12 hours ago, jaapv said:

Y

We may as well accept that photography is a technological process that will always alter reality. 

I

 

DON'T TAKE MY REMOVE AWAY!   

[Stephen.s1]

On a more serious note: I as "artist" can and should do anything he wants with his creation.  

[LocalHero1953]

This is the first time I have thought about what my own practice should be for AI. As a first draft (and I may change my mind!), I should:

• state if I have changed something that was originally in the scene i.e. adding/removing/replacing elements*.
• not state changes that are a consequence of taking the photo (noise, dust, poor exposure, focusing).
• not state changes that clarify the original image (masking to adjust lighting).

 

*Edit. This will require me to mention things I haven't mentioned before. In photographing play rehearsals, I often remove light switches, exit signs, water bottles etc that distract from the background. I will mention this in future.

Edited September 17 by LocalHero1953

[LocalHero1953]

5 minutes ago, Stephen.s1 said:

On a more serious note: I as "artist" can and should do anything he wants with his creation.  

Of course, if you set your own standards for satisfying yourself, then you can do what you like.

If you are reporting factual information, looking for 'likes' on social media, or aiming for a qualification in photography, then that's a different matter.

Edited September 17 by LocalHero1953

[Advertisement]

Hallo,

Look here: Interesting Leica gear

[Dazzajl]

As an aside, the fact that AI can create ‘art’ while I’m still cleaning and doing laundry means that we have everything the wrong way around! 

[pgh]

4 hours ago, 01maciel said:

That is a different matter; political discussions are out of place here. So please, don't and stop it.

 

It's not though - there are direct political implications if this technology is widely adopted and accepted amongst people looking at photorealistic images. It cannot be separated, and it is incumbent on viewers and photographers to come together to redefine the medium in the age of this technology - or, if not, no one trusts anything they see anymore, which leads to widespread cynicism which again, has direct political results as @hansvons notes. These things are not unrelated.