Artist Refuses Prize After His AI Image Wins at Top Photo Contest (petapixel.com) 108
An anonymous reader shares a report: A photographer has stirred up fresh controversy and debate after his artificial intelligence (AI) image won first prize at one of the world's most prestigious photography competitions. He has since declined to accept the prize while the contest has remained silent on the matter. Berlin-based "photomedia artist" Boris Eldagsen participated this year in the World Photography Organization's Sony World Photography Awards, a leading photo contest that offers prizes that include $5,000 cash, Sony camera equipment, a trip to London for the awards ceremony, and/or worldwide publicity through a book and exhibition. Eldagsen submitted an image titled THE ELECTRICIAN to the Creative category of the 2023 Open competition. It picture appears to be a portrait of two women captured with a photographic process from the early days of photography.
Imagine (Score:3, Insightful)
If you can imagine it, AI can build it for you. No real talent (besides imagination) required.
This will change everything.
Re:Imagine (Score:5, Insightful)
Re: Imagine (Score:2)
I mean - thatâ(TM)s the point. AI has got to a point that a sufficiently skilled prompt whisperer can produce images that can be confused for with the best photographs in the world.
Re:Imagine (Score:5, Interesting)
Asking how to define the difference between digital photography and an AI generated "photo".
Detecting the difference is potentially difficult, but defining it in negative terms is easy. If there's not a real subject involved, an image of whom is being produced by light bouncing off of it and then interacting with a sensor, film or similar, then it's not a photograph.
Re: (Score:1)
That is a fairly good attempt and doges many of the tricky cases like an AI driven red eye correction filter but still not simple. AI is really just regurgitating training input run through a digital blender and there were probably MANY real subjects with images produced by light bouncing off of them and hitting a sensor.
Re: (Score:3)
That is a fairly good attempt and doges many of the tricky cases like an AI driven red eye correction filter but still not simple. AI is really just regurgitating training input run through a digital blender and there were probably MANY real subjects with images produced by light bouncing off of them and hitting a sensor.
Your highly reductive argument is the same thing as claiming that "cranberry-flavored juice cocktail" is the same substance as actual cranberry juice.
Yes, we get that both substances are the outcome of harvest and processing of raw materials, blending, pasteurization, filtering, etc. But the fact that both were processed to get them into their final form doesn't somehow automatically reduce and wave-off the difference between a substance that IS cranberry juice and a substance that contains 5% cranberry ju
Re: (Score:2)
"Yes, we get that both substances are the outcome of harvest and processing of raw materials, blending, pasteurization, filtering, etc. But the fact that both were processed to get them into their final form doesn't somehow automatically reduce and wave-off the difference between a substance that IS cranberry juice"
That would be an apt analogy if the substance which IS cranberry juice where 15% not at all cranberry juice due to changes in the processing and the cranberry-flavored juice cocktail was also com
Re: (Score:2)
"Yes, we get that both substances are the outcome of harvest and processing of raw materials, blending, pasteurization, filtering, etc. But the fact that both were processed to get them into their final form doesn't somehow automatically reduce and wave-off the difference between a substance that IS cranberry juice"
That would be an apt analogy if the substance which IS cranberry juice where 15% not at all cranberry juice due to changes in the processing and the cranberry-flavored juice cocktail was also composed entirely of cranberry juice and similar juices because there is nothing in the training OTHER than photos and artwork in the training data
This seems like the other post where you tried to reduce/equate photography to requiring film.
"Juice" is not the same thing as "natural liquid".
Nobody is holding a cranberry over a bucket and squeezing out the "juice". Juicing something doesn't mean extracting only the parts of it that were already in liquid form in its natural state. Cranberries and numerous other fruits/vegetables do not contain enough liquid enough to be commercially viable that way.
"XYZ juice" means "a liquid derived from processing XYZ
Re: (Score:2)
"This seems like the other post where you tried to reduce/equate photography to requiring film."
There was no post where I tried to reduce or equate photography to requiring film. But failing to understand that is more or less in line with the rest of your commentary.
""Juice" is not the same thing as "natural liquid".
Nobody is holding a cranberry over a bucket and squeezing out the "juice". Juicing something doesn't mean extracting only the parts of it that were already in liquid form in its natural state. C
Re: (Score:2)
"This seems like the other post where you tried to reduce/equate photography to requiring film."
There was no post where I tried to reduce or equate photography to requiring film. But failing to understand that is more or less in line with the rest of your commentary.
I apologize. It appears I mistook Shaitan for the Archangel Michael [slashdot.org].
Which is a hilariously accurate summary of my life.
I would point out, however, that my brain hardware's limitations in perfectly preserving distinctions between different usernames (and its tendency to use shortcuts like meta-categories whereby the name of a demon and the name of an angel are tagged as related and thus more prone to transposition errors like Tuttle and Buttle) does not mean that there is in fact no distinction.
Did you really double down on your broken analogy without even bothering to take the time to comprehend why it was broken?
"A photograph is an image of a person/place/thing, composed of the light (or lack of light) captured by a light-capturing device within a particular spacetime reference frame."
That definition fails when you perform digital editing.
It does not f
Re: (Score:2)
"In the case of an AI system the content is pieces of data from thousands or even millions of photographs which are then regurgitated piecemail into a novel combination. In other words it is 100% cranberry, the bits just didn't come from this cranberry or any one cranberry.
I believe you are focusing more on questioning the output, where my definitions/analogues focus more on questioning the input. This leads you to focus on the forest and (from my perspective) undervalue the trees."
On the contrary, I'm focu
Re: (Score:2)
"In the case of an AI system the content is pieces of data from thousands or even millions of photographs which are then regurgitated piecemail into a novel combination. In other words it is 100% cranberry, the bits just didn't come from this cranberry or any one cranberry.
I believe you are focusing more on questioning the output, where my definitions/analogues focus more on questioning the input. This leads you to focus on the forest and (from my perspective) undervalue the trees."
On the contrary, I'm focused on a technically correct definition. You seem to be focused more on the spirit of the thing.
Reading your entire reply, I can more clearly see where you're coming from.
Though in the above passage I would have said the opposite -- you are focused on the vernacular spirit of the thing, and I am focused on trying to statutorily lay out what would be a correct (falsifiable) technical definition of the word "photograph".
Since you are focused on the modern everyday spirit of the thing, you are able to accept all sorts of images, produced from a variety of methods, depicting a variety of real/unreal subje
Re: (Score:2)
"Since you are focused on the modern everyday spirit of the thing, you are able to accept all sorts of images, produced from a variety of methods, depicting a variety of real/unreal subjects, as "photographs"; and furthermore you have yet to give clearly limited technical criteria for what defines a "photograph" and differentiates it from an "image" "painting" "rendering" "collage" and so on.
Since I am focused on a more technically-defined, etymological understanding of "photograph", I have a much narrower
Re: (Score:2)
That doesn't change the definition, it only makes it harder to detect the real thing.
So far, the more complex the output, the more signs there are that it's AI generated.
Re: (Score:2)
I'm not sure that part matters (the training data) - photorealistic images by a painter, digital artist *or* AI doesn't a photograph make, at least in most colloquial use I've encountered.
I'd say similarly there's been a lot of discussion in the past about composites vs photographs, and when or if they become "digital art" vs photographs. This seems more like an extension of that.
Re: (Score:2)
"I'm not sure that part matters (the training data) - photorealistic images by a painter, digital artist *or* AI doesn't a photograph make"
There is a very distinct difference between your first two examples and your third. The first two are making an original creation based on their memory and interpretation of photos rather than regurgitating the actual data from the original.
If I have a million people paint a starry night there won't even be an identical blue (even if told the ratios to mix!), no identica
Re: (Score:2)
I just think there's no reason to think painters aren't inherently doing the same thing the AI is - they're influenced by any art they've seen in the past. They are taking very small bits of style and what they think a window would look like in this situation, and mixing it together and making a somewhat original output.
Lets take fantasy content, something there is no "original" for the painter to see. Their representation of a dragon is just a remix of all the other drawn or painted dragons they've seen.
Wh
Re: (Score:2)
"I just think there's no reason to think painters aren't inherently doing the same thing the AI is - they're influenced by any art they've seen in the past. They are taking very small bits of style and what they think a window would look like in this situation, and mixing it together and making a somewhat original output."
And yet the training photos have been identified from AI output because the AI incorporated bit by bit identical chunks.
Yes, the human painter is influenced by everything they've seen befo
Re: (Score:2)
And yet the training photos have been identified from AI output because the AI incorporated bit by bit identical chunks
I'd love to see the information around that. How big is this chunk? Because there's a threshold where obviously it's not an "identical chunk" by any reasonable analysis, yet technically it is an identical chunk. What I mean is, if it's one pixel - cannot be a copy / taken chunk by definition really. Or it loses meaning and plausibly could be "taken" from about everything. But in high res im
Re: (Score:2)
In some cases entire images... https://www.vice.com/en/article/m7gznn/ai-spits-out-exact-copies-of-training-images-real-people-logos-researchers-find
Re: (Score:2)
"The painter has self-awareness, tastes, preferences independent of ANY input
All of that is basically random transforms adjusted by external inputs (life experience)."
I'm sorry but that is a statement of faith not science. This is essentially the same universal function we plug into evolution on the notion that eventually it can't accidentally arrive at any possible answer and it has had billions of years to do so. We aren't building these systems to operate over multi-billion year lifespans. We also no lon
Re: (Score:2)
"This AI isn't outputting plaster casts, those are exact copies of a whole sculpture."
An artist in this case isn't an end user using the AI but the developer applying AI to create a solution like this. AI is a programming tool. I'm not talking about code suggestion AI's I mean the neural net frameworks themselves. They are built to logically conform around the problem in a mechanical manner metaphorically similar to the plaster. The problem is the thing being cast, it's resistance the feedback and the gravi
Re: (Score:2)
I just think that at a fundamental level, humans are doing the same sort of thing computer systems are, just that it's more complex. In terms of "same thing" I mean information processing, not some very specific set of steps. Do you think there's a important difference between a traditional chess program and alphachess? They both play chess, often better than humans can, but one is programmed by humans and one is AI self taught. But at the basic level, they are inherently computer programs running to play c
Re: (Score:2)
I'm sorry but that is a statement of faith not science.
I guess you could phrase it that way. Do you think the brain is not doing information processing via physical processes?
This is essentially the same universal function we plug into evolution on the notion that eventually it can't accidentally arrive at any possible answer and it has had billions of years to do so. We aren't building these systems to operate over multi-billion year lifespans.
I don't really understand what you're saying here, or the rel
Re: (Score:2)
I think we are more or less reimplementing a major part of the machine we are driving. It really is amazing and can do all sorts of things... perhaps one day these kind of systems could be wired so that the wetware in our brains can integrate in them and expand the capacity of our machine. But we are not replicating the driver nor are we really trying to... because we want to be the driver and in control.
To answer your direct question I think a traditional chess program speeds up the way we tend to consciou
Re: (Score:2)
"I might be more materialistic than you, but I'll say the main difference between AI and a painter in the creativity area is just that humans are still more complex math we don't understand as well yet. At it's base, it's still just physics happening."
Maybe, maybe not. If as we are starting to believe there is a quantum information aspect then the physics vs other factor stops really having much meaning anyway.
I'm not saying we couldn't create a human-like self-aware intelligence. Nobody is really trying to
Re:Imagine (Score:5, Insightful)
Photography should require film then?
Don't be silly. It should capture a moment of reality, by whatever means.
Not synthesize anything. That is the domain of other arts.
Re: (Score:1)
Ah. Can it run then trough pixel editing?
Frankly, we are not well at all with photography for a good while already, since the advent of cheap pixel manipulations.
This is just another level of faking images. Photography is too big of cultural achievement, to be lost like that.
Perhaps, there should be reservations with strict terms to the processes acceptable.
Observing local gathering of enthusiast photographers in e-space, I see less and less of production to have qualities of original art form. Not good.
Fas
Re:Imagine (Score:4, Insightful)
I never made the transition for film to digital in terms of photography as a hobby. I obviously still take pictures, but they are mostly of a documentary nature now, images of family events and such not much 'art' for its own sake, even if I still think about composition, and practice some of those skills.
In my memory at no point was photography without its manipulations. In the film world photographers were never just capturing an image, they were always very intentional about the selection of IOS speeds, and depth stops - not just to calculate a proper exposure but to create specific effects, stop action, blur motion, limit the focal range, maximize the focal range, create more intense contrast or less. People would bracket exposures one stop each way, or maybe 1/32s of time one way or the other, and chose the one they liked later. Next we went to the dark room where we again played with the exposure, maybe we soften the edges of images with vignetting, possibly softened the entire image by having the enlarger just slightly out of focus, perhaps we go really wild and did double exposures to 'thicken up forest' or do silly things like have someone shake their own hand for examples. A lot of people liked to paint negatives... Color was always subjective, I don't know anyone making prints that just went with whatever the diffuser and test strip indicated for color filters. Everyone played around and for the mast part even folks going for a "natural" or "true" quality, were still being less than scientific.
Glossy, matte, finishes, various washes...
In earlier eras of photography it was working around limits of the technology, maybe you took a picture in broad daylight, because you needed the light but staged a nighttime scene..
Photography has always had an element of technical manipulation to create the image we 'want' which was not the image we simply saw most of the time. In many respects that is what made it art. Its like a lot of other formerly human endeavors though right now; when is it not the artist but the machine that is doing the created, how many driver assist tools like stability control can you put on an F1 car before a race is no longer about driver skill? Plus the more philosophical question does it matter?
Re: (Score:2)
We do address it differently, when speaking about value in photography, as a form. Playfulness and flexibility are not problem, until one substitutes it with pixel painting altogether. Or now - by pixel generator. With the end goal set being pleasability, instead of photo-graphical *representation*. Everybody now is invited to have a say what is good photography, hierarchies are washed out. Like everybody now is photographer, because of mobile-embedded camera. I just have had about a decade of toying with f
Re: (Score:2)
Re:Imagine (Score:4, Interesting)
Photography should require film then?
Yes. It should. When its a contest with a prize. Its easy to drive from Hopkinton to Boston, but we still insist that the winner of the Boston Marathon has to run the whole way.
Re: (Score:2)
Well that's silly to me. Maybe you are going to be pedantic on what a photograph is, but whether I use a digital camera or a 35mm, I can get both printed to the same card stock and picked up at walgreens. Saying the digital picture of my cat is not a photograph but the 35mm of the same cat in the same chair IS a photograph is, for lack of a better word, silly.
Either piece of technology is capable of producing a photograph of my cat sitting in a chair.
With respect to the article, AI creating a photograph of
Re: (Score:2)
I'm not being pedantic about what a photograph is. I'm being pedantic about what a contest is.
Re: (Score:2)
Okay that's a valid point!
Re: (Score:2)
And yes, I realise that every digital photograph is in some way artificial and you can add filters or whatnot to any kind of photography. So there might be a grey area. Still, his image was not a photograph and his treatment by the organisation shows how the world is not yet ready for AI.
Re:Imagine (Score:4, Insightful)
Have you actually tried to use AI to generate pictures? Because in my experience, you put in some pretty general terms, are amazed out the output, and retroactively conclude, "Oh yeah, that's exactly what I had in mind." In fact, because my actual intent wasn't even really understood by me, nearly any output fit the bill.
Here you can see the actual piece - "The Electrician", by Boris Eldagsen: https://www.engadget.com/germa... [engadget.com]
If I typed until my fingers were stubs, I couldn't do that.
Re: (Score:3)
It's one of several [eldagsen.com] images. This is what he says about them:
[...]These images were imagined by language and re-edited more between 20 to 40 times through AI image generators, combining “inpainting”, “outpainting” and “prompt whispering” techniques.
Just as photography replaced painting in the reproduction of reality, AI will replace photography. Don’t be afraid of the future. [...]
Very angst, many pain (Score:2)
Re: Imagine (Score:1)
The whole point of photography is to capture a moment in real life no? Certainly we can debate the use of photoshop in touching up/editing a picture but at the core of it all there was indeed a real-life picture that was taken by the photographer themselves.
Granted I can understand that these are the guyâ(TM)s words, not yours but this just sounds like one of those vague and grandiose statements a lot of self-proclaimed visionaries make to sell products.
Re: (Score:2)
Well, I wouldn't describe they whole point of photography that way. Given that it can be abstract, timeless, staged, misleading, doctored, or HDR'd into to being more cartoon that real, I would say purpose of photography is at best undefined. More often than not, the goal isn't truth... even if the scene is "real", the result is idealized.
Real life mostly doesn't happen in the golden hour. It has blemishes and squinting, your good side isn't always the visible one, and the sun is in your eyes.
Re: (Score:2)
The whole point of photography is to capture a moment in real life no?
No. Look for other post here on this story for deeper thoughts on the subject but I would say the point of photography was always to capture an artists conception of how they wanted the subject to appear.
Re: Imagine (Score:1)
Fair, but in the case of AI-generated photographs the subjects (unless explicitly defined) never really existed to begin with. At best they are drawn from training data that contained images that the AI draws from. At least with photographs it works off a base image, regardless of how much you put it through photoshop.
In that regard, would it not be more akin to a very photorealistic CGI? Which in itself I consider an art form but one wholly distinct from photography.
Re: (Score:2)
Re:Imagine (Score:4, Interesting)
Okay, I'll bite. Using any of the numerous, frequently free, image generators out there, with no postprocessing effort, create an image for me of a horse riding on a man, talking to an RJ Palmer-style Charizard, standing next to a sign that says "Welcome to Mississippi".
I'll wait.
Re: (Score:1)
Re: Imagine (Score:1)
It's also going to get much easier to do. Midjourney recently released a tool that will allow for img-to-text so you can break an image down into prompts to recreate it. Even if you are not planning on making a 1:1 recreation of an image, it still takes a lot of the guesswork out of it. And no doubt it will get better especially when paired with prompt optimization tools. In other words, the automation is already getting automated.
Really, the irony is that a lot of arguments being leveraged against regular
Re: (Score:2)
AUTOMATIC1111/SD has had *two* versions of image-to-text for ages. It's NOT all-that. You may get images kinda-like your starting image, but not that close, and all still with your typical range of AI flaws. Because at the end of the day, new images are still generated by prompts, to the best of the AI's ability. And said "best of the AI's ability" is limited.
The biggest game changer in the history of AUTOMATIC1111 / SD has been, by contrast, ControlNet (which MidJourney has nothing similar). *That* is
Re: (Score:3)
It literally does not, which is why I told you to do that.
It constantly astounds me the people who have no experience with AI art who adamantly insist to those who use the tools every day that they can just do everything, right out of the box, with no learning curve and no postproessing.
That is just NOT the reality.
Re: (Score:3)
Just to elaborate on what will happen if you try:
* It'll have a man riding a horse instead of a horse riding a man, because it sees "riding" and "man" together and can't conceive of any other arrangement (assuming it actually gives you both a man and a horse).
* The very mention of "Charizard" will turn your whole image cartoony, even if you try to stop it, and RJ Palmer's "photorealistic pokemon" works aren't trained into the model, so if you want his style, you have to custom-train, and that can be
Re: (Score:2)
No, it can't. I tried exactly that, on several of the "AI" image generation sites. I gave an "AI" a very detailed description of what I want, and the AI failed miserably. Took me more than 4 hours to do all that. I made a nearly perfect photoshop of what I wanted in about 30 minutes.
Re: (Score:2)
This will change everything.
It will change nothing. There are already million of high quality photos/drawings made every year with probably an average of 20 views each, so now we will have 2 million more, so what? Artists will still get famous at random (luck and social skills). I don't expect artists to be fired either, they will produce more.
Ai - The new Steroids (Score:5, Insightful)
You can't just use AI and suddenly be top tier, just like you can't just take steroids and suddenly be a competitive bodybuilder... but given equivalent aptitude and willingness to really do the work, the AI user will eventually have the upper hand.
Re: (Score:3)
This.
This is something most people who haven't spent much time with AI tools don't understand. They just think, "you type in whatever you want, and it does that". In reality, the AI tools have some amazing strengths, but also amazing weaknesses. Working human + AI together, you can achieve amazing things, but doing so requires lots of experience and not insignificant labour (though less than without AI tools, in most cases at least).
Re: (Score:3)
If you generate enough images with the same prompt, eventually the software spits out one that's good enough to win a competition without any retouching.
Is that the genius of hard work, or just perl-type "true" laziness?
If you want to achieve a very specific goal with the software then that takes a lot of work, but getting a really amazing image out doesn't necessarily.
Re: Ai - The new Steroids (Score:1)
I think this is more reflective of the fact that the average consumer of art/media just does not have a discerning eye for quality, let alone an understanding of what goes into photography. It is all just pretty pictures to them and not much else.
Re: (Score:2)
Well, yeah. I mean, who is the art for? If your only goal is to reach experts, your potential audience is kind of small. And there is art like that, of course, but there's way more that is meant to appeal to a broader audience.
Re: (Score:2)
"If you generate enough images with the same prompt, eventually the software spits out one that's good enough to win a competition without any retouching."
It most definitely does not. Punch in "A photo of a horse riding on a man, talking RJ Palmer-style Charizard, next to a sign that says "Welcome to Mississippi"."
I'll wait.
SD doesn't friggin' understand "A red box on a blue box", and you think it just generate anything if you click 'generate' enough? Sure, in the same way that a million monkeys will even
Re: (Score:2)
Re: (Score:2)
Welcome to the brave new world.
guy's got dignity (Score:2)
Because focus on one thing at a time (Score:3)
good on him to refuse these attractive prizes. The picture feels a bit AI-generated to me, not sure if prior knowledge of that information is clouding my judgement though
Look at the direction that the two eyes of the frontmost person are pointing: they're not both looking in the same direction.
Also of note: The reflections from the frontmost person's clavicles (on the chest, just under the throat) are not symmetric and the gap between is not in the middle of the chest.
The ear of the rearmost person is way too far back on the head, so that the junction between the ear and the head is over the horizon of the curve of her face.
The cheekbones of the frontmost person are not the
Re: (Score:2)
Good call on the clavicle asymmetry.
The spacing on the fingers of the hand that's on her shoulder leapt out at me as wrong. Also the fingernails.
Re: (Score:2)
I'm much more impressed by this one:
https://www.eldagsen.com/pseud... [eldagsen.com]
The folds under his chin are a bit weird, the hands seem a bit big, but apart from that...
Re: (Score:2)
There's also something off on the back woman's hands, they don't look right at all. I've seen that with other AI created images, it gets the hands and eyes wrong.
Re: (Score:2)
I came here to say exactly that, well, less eloquently, for sure, however the image looks bad in so many ways.
The person behind is creepy, looks like a serial killer who's looking for her next fix, thinking "what if I suddenly choke this young lady?".
Her left hand fingers are all weird. And so on.
Re: (Score:2)
Re: (Score:2)
Dress drapes merging into the arm itself in anything but natural way.
All in all, this thing is ugly disgrace to what human were considering an artistry.
This winning is also insult to the qualifications of those judging.
Seriously? (Score:1)
Not controversial. (Score:3)
He should simply be disqualified. It's a photography contest where no photography was performed. It would be more controversial if he entered a generic art contest.
I'm generally in favour of image manipulation, Photoshopping is an integral part of photography just like the darkroom was integral in the days of film, but unless he fed an original image from a camera into some algorithm and used AI to manipulate it, the image has nothing to do with photography.
Re:Not controversial. (Score:5, Insightful)
That's why he refused the prize:
> Eldagsen subsequently refused the award, saying "AI is not photography. I applied [...] to find out if the competitions are prepared for AI images to enter. They are not."
The controversy isn't that he entered (Score:2)
Re: (Score:3)
Or they did catch it and knew the whole time it was AI, but they wanted to get exposure to this issue and awarded him so that it would all come out. Strike while the AI iron is hot. If they didn't do this then some other contest would and would get all the exposure. I did recall it happening earlier this year somewhere else also.
Re: (Score:2)
Occam's Razor says no.
If they didn't do this then some other contest would and would get all the exposure.
These photography contests are dime a dozen and usually quite richly entered. They don't need "exposure".
Re: (Score:2)
it's that the judges not only didn't catch him but awarded him.
Again, not controversial. The art world has a rich history of fraudulent entries getting awards for things. This is just business as usual.
Re: (Score:2)
He should simply be disqualified.
Agreed. His entry violated the rules. Since his entry was subterfuge, and he is benefitting from the attention of refusing the award, he should probably be banned from future contests.
It's the only ethical thing to do. (Score:2)
It's the only ethical thing to do, as he didn't create the work. His computer did. He will be known among artists as so lacking in creativity that he had to have a machine help him. We're not looking at his work, but the work of a very complicated mathematical model.
Right now, using AI art is controversial among artists, to say the least. A lot of them absolutely hate it. It's not merely that it has the potential to displace working artists, but the fact that AI removes almost all of the artist's cr
Re:It's the only ethical thing to do. (Score:5, Insightful)
No feeling, no emotion - it is simply an image which mimics what others have done. It has no soul.
Yet somehow the jury picked it as a winner.
Saying there is no art involved in a picture like this is disingenuous. It takes a lot of work, it requires a specific concept and it requires creating the many different part of the image, the style, etc. Of course it's creative.
If you as a photographer take a picture of tree, you didn't create the tree, you didn't create the sky, you didn't create the lens, you didn't create the sensor and you didn't create the image processing code.
So what exactly is the art in the photograph then? Well it's mostly about the idea, the composition, technical execution and finding the right subject. Not that different from the AI one.
Of course the artist is right that a photography contest isn't the place for AI work.
Re: (Score:2)
Of course it's creative.
According to his web page:
"Using the visual language of the 1940s, Boris Eldagsen produces his images as fake memories of a past, that never existed, that no-one photographed. These images were imagined by language and re-edited more between 20 to 40 times through AI image generators, combining “inpainting”, “outpainting” and “prompt whispering” techniques."
So... not at all easy to do.
It's not photography though, and this was a photography competition.
Re: (Score:1)
It's the only ethical thing to do, as he didn't create the work. His computer did. He will be known among artists as so lacking in creativity that he had to have a machine help him. We're not looking at his work, but the work of a very complicated mathematical model.
Right now, using AI art is controversial among artists, to say the least. A lot of them absolutely hate it. It's not merely that it has the potential to displace working artists, but the fact that AI removes almost all of the artist's creativity.
AI generated imagery is not art. It is a manufactured image, much like CGI. The key difference between AI imagery and other digitally created artwork is that with AI, the "artist" specifies what is created, but not how it is created. An artist uses the brush or pencil to determine not just what is created, but how it is created, rather than just choosing one of several options presented by the AI model. With every brushstroke, every line, the artist tells us about himself, his perception, his understanding of the world, his experience. AI art contains none of that. No feeling, no emotion - it is simply an image which mimics what others have done. It has no soul.
AI doesn't enhance an artist's creativity, it removes it.
Define "art".
Preferably in an operational fashion, that is, describe a process by which I can decide whether an image is "art" or not, without knowing whether computer or human made it.
I'll wait.
Re: (Score:2)
Heck, we can't even define art when humans do it - think of inkers, animation artists who did the in between frames, and more - the banana on a wall, or the urinal on a stand. Art is such a generic term that anything can be art IMHO.
Re: (Score:2)
The best answer I've heard is that art, at its core, is about communication, so anything done with the intent to convey a concept is art, often that concept is simply aesthetic beauty.
Unfortunately this isn't a very satisfying definition as it is so broad to include things like smiling at someone, but it does put some limits on it.
Say you found a cool rock and put it on display, the rock itself is not art, but the presentation of it is.
Re: (Score:2)
What differentiates art from every other human endeavor is that it is creative activity conducted for the purpose of creation, without any considerations of practical value.
It is impossible to identify the difference between human creativity and computer generated imagery by aesthetics alone; even though mindless, geological processes created the mountains, we would not call them art even though they're beautiful.
What this guy did was neither art nor photography - it wasn't photography because it wasn'
Re:It's the only ethical thing to do. (Score:4, Insightful)
I'm going to disagree. It is art, though it is not photography.
Re: (Score:2)
It is not quite that simple.
From TFA:
“The work SWPA [Sony World Photography Awards] has chosen is the result of a complex interplay of prompt engineering, inpainting, and outpainting that draws on my wealth of photographic knowledge. For me, working with AI image generators is a co-creation, in which I am the director. It is not about pressing a button – and done it is. It is about exploring the complexity of this process, starting with refining text prompts, then developing a complex workflow, and mixing various platforms and techniques. The more you create such a workflow and define parameters, the higher your creative part becomes.”
Eldagsen says that he calls his work “images” and not “photographs” since they are “synthetically produced, using ‘the photographic’ as a visual language.” He also says that he is trying to bring this distinction to the forefront in the photo contest industry so that separate awards can be created for AI images.
Re: (Score:2)
It's the only ethical thing to do, as he didn't create the work. His computer did.
That's like saying Vermeer didn't create his work, his paints and brushes did.
He didn't just wake up one morning and find a new jpg file on his desktop.
Re: (Score:2)
Re: (Score:2)
How so? Apparently the process to make this image took many hours of the artist working on it to get it to express exactly what he wanted, i.e. the emotive content and artistic style. It's not fundamentally different from him doing the same work in Photoshop, or with a paintbrush. It's just not photography, so he was right to refuse the prize.
I don't like aspects of AI image generation (e.g. copyrighted training images resulting in a rash of garbage imitating an existing artist's style, cash-grabbing by com
Re: It's the only ethical thing to do. (Score:2)
Re: (Score:2)
You also can't put into brushstrokes what you are feeling that closely. It's not magic, dude :) I sometimes do agonize for hours over a single detail, but this is neither universal nor a requirement for great art. As a rather famous example, Vincent Van Gogh is said to have finished a new painting every 36 hours on average during his productive periods. Some people might even say that spontaneity and serendipity are the key to artistic expression, and overthinking is actively detrimental.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
s/computer/camera and you've got early perceptions of photography as 'art'. This tempest in a teapot demonstrates a need to add an AI media or category.
Its still the hands.. (Score:3)
At first glance there's just that tiny bit of "something you cant put your finger on"
But look on image right - that hand on her chest .. the angle is all wrong for it to be hers and it's a left hand.. so ... yeah like still if you're not looking for it you just get that bit of "it's a bit off" without the WHY registering.
One day, AI is going to get hands/arms right and oh boy are we going to be in trouble...
Re: (Score:2)
And of course like most CGI, the lighting doesn't really make much sense, unless we're to believe someone set up some rather unusual lighting for this picture.
Re: (Score:2)
Perhaps the most interesting aspect... (Score:3)
... of this story is that the only reaction of the organisers (the World Photography Organisation) to the artist's (Boris Eldagson) attempt to initiate a discussion about artificial intelligence in photography awards was that they quietly removed all traces of his participation.
Re: (Score:2)
Re: (Score:2)
That's doesn't do the case justice. His intention obviously was and is that, other than himself, there will be more and more people entering AI generated material for photography awards without saying so afterwards, while the World Photography Organisation seems to be intent on ignoring that.
Re: (Score:2)