Photography AI as art

[faberryman]

Imagine a gallery full of Thomas Kinkade paintings. If there was one AI generated painting on the walls I'd likely see it as refreshingly creative in comparison. It's a low bar and don't know if there's a photography equivalent, but it's an interesting thought experiment.

I see photographs in the Gallery that would give Thomas Kincaid a run for his money for the Most Needs Improvement Award.

 

[nmp]

Imagine a gallery full of Thomas Kinkade paintings. If there was one AI generated painting on the walls I'd likely see it as refreshingly creative in comparison. It's a low bar and don't know if there's a photography equivalent, but it's an interesting thought experiment.

Gets us right back into the 'what is art' debate.

Yeah Kincaid came to my mind when seeing AI generated images on this thread.

AI Luminous Landscapes

AI Luminous Landscapes

forum.luminous-landscape.com

The way I look at is as long as there is the tool, people will use it. They will get better at it and be more creative and some of it will be art. People who are practicing conventional photography have nothing to fear - their art will become even more niche and higher in demand as more and more new entrant will take up this new form of creation.

Just keep doing what you are doing well....

:Niranjan.

 

[MurrayMinchin]

Have we defined AI photography yet?

Wouldn't it be images where the program comes up with images in a completely autonomous way...if that's even possible, as somebody had to program it in the first place?

I use an "AI" app (ON1NoNoise) to help smooth out high ISO digital camera blotchiness when taking bird/wildlife photos at fast shutter speeds in low light. I start with the slider at minimum and slowly apply the app until things get smooth, but if you apply it to the max, it looks like Max Headroom. I also control where it's applied.

I like VinceinMT's suggestion that photographs need to made with light, out in the messy real world.

 

[snusmumriken]

Yes, it's that simple.

And a bit depressing that we're still having this "conversation"—although its "How many angels can stand on the head of a pin" aspect doesn't make it worthy of that term—a hundred years after Duchamp's ready-made, where in the "Is _____ art?" we've just replaced "a urinal" by "AI photography".

Most important issues related to the use of AI will deal with ethics, not aesthetics.

No it isn’t that simple! As I said in post #2, an image could have merit because it’s likeable to me, likeable to lots of people, likeable enough to own and live with, worth preserving for the nation, or a good investment. An AI-generated image could be any of those things.

Some folk, like the OP, feel that AI is cheating. I don’t, because I don’t see art as a competition with rules. My appreciation of things is not dictated by anyone else. This morning I picked up a fossil in the field next door because I liked the look of it. It isn’t art: there was no artist. But it will grace my bookshelf, and I will appreciate it whenever I look at it. What it does for me matters far more than who or what made it. Similarly, if I like an AI image enough I may hang it on my wall.

Unfortunately (or fortunately) viewers are unpredictable and untrustworthy. A lot of people would say my fossil is junk. Probably my descendants will chuck it away. Maybe I will tire of it myself. The same applies to images. A lot of people claim to have made art, but no-one likes it. Sometimes its time comes later, but it may not come at all. Sometimes its only time is now, and very brief. The same will apply to AI images as it does to others.

 

[nmp]

Have we defined AI photography yet?

Wouldn't it be images where the program comes up with images in a completely autonomous way...if that's even possible, as somebody had to program it in the first place?

I use an "AI" app (ON1NoNoise) to help smooth out high ISO digital camera blotchiness when taking bird/wildlife photos at fast shutter speeds in low light. I start with the slider at minimum and slowly apply the app until things get smooth, but if you apply it to the max, it looks like Max Headroom. I also control where it's applied.

I like VinceinMT's suggestion that photographs need to made with light, out in the messy real world.

I think most when they use the term "AI" mean Generative AI - meaning creating the images out of thin air - not what you do to clean-up your digitally captured image - so you are safe....

:Niranjan.

 

[Alan Edward Klein]

I think most when they use the term "AI" mean Generative AI - meaning creating the images out of thin air - not what you do to clean-up your digitally captured image - so you are safe....

:Niranjan.

You need more than thin air to capture a picture.

 

[Pieter12]

I think most when they use the term "AI" mean Generative AI - meaning creating the images out of thin air - not what you do to clean-up your digitally captured image - so you are safe....

:Niranjan.

As far as I know, AI cannot generate images out of thin air-- to work it needs to "learn" (steal, really) from elements that already exist.

 

[koraks]

You need more than thin air to capture a picture.

In cases of very thin air, one could consider using a UV filter to cut down haze. At least in broad daylight.

 

[warden]

As far as I know, AI cannot generate images out of thin air-- to work it needs to "learn" (steal, really) from elements that already exist.

And the more AI is used the more it will tend to steal from itself as it inbreeds. The tech will need to make sure it isn't using old AI to make new AI or quality will decrease over time.

 

[nmp]

As far as I know, AI cannot generate images out of thin air-- to work it needs to "learn" (steal, really) from elements that already exist.

Thin air meaning the person creating it does not have to feed images - they can just say draw me a dinosaur watching Jurassic Park in a the theater. And the AI program draws. Thin air of the person's imagination, is what I meant.

:Niranjan.

 

[nmp]

You need more than thin air to capture a picture.

What capture?

 

[nmp]

And the more AI is used the more it will tend to steal from itself as it inbreeds. The tech will need to make sure it isn't using old AI to make new AI or quality will decrease over time.

Spooky....

 

[Sirius Glass]

I should have added "economics."

I was at a two-day conference a couple weeks ago about radio and podcast, in which the main subject was AI. It went from AI tools to help you create your podcast (there are some really good ones) to a radio station, RadioGPT (won't give the link, but you can Google it), in which everything, from the music choices to the voices doing news, traffic, and, at times, commentary, are all AI generated. Some voices still sound a tad robotic, but others, especially those cloned from actual voices, are sounding more and more natural, and will totally sound so in a couple of years. Company who built this technology has already started to sell it to actual radio stations in the US.

So, main question associated will all this during the conference wasn't "Is it radio?". Main questions were "is it ethical?" (or when does it stop being ethical, which is more complicated) and "How many people will lose their job?".

So yeah, it will be photography because the main use of AI won't be in art—I suspect number of artists doing fully AI-generated photographic artworks and being successful at it will be minimal. It will be in fashion and advertisement—domains where the notion of ethics can already be at times a bit loose—, and the people losing their jobs will be photographers.

Just because one can do something does not make it right or even ethical. One can enslave another person because they can, but that is neither right nor ethical.

 

[Sirius Glass]

I agree. But it's not photography.

We agree on something!

 

[Sirius Glass]

I think most when they use the term "AI" mean Generative AI - meaning creating the images out of thin air - not what you do to clean-up your digitally captured image - so you are safe....

:Niranjan.

No, not out of thin air, but out of input data.

 

[nmp]

No, not out of thin air, but out of input data.

Again I don't know what the problem is. The person who is USING AI does not input data. That guy sitting behind your terminal does that. You provide the thin air of your imagination. AI term itself is a bit of a misnomer - it's more like machine learning on steroid. Of course, you have to feed the machine to learn on.

:Niranjan.

 

[Sirius Glass]

Again I don't know what the problem is. The person who is USING AI does not input data. That guy sitting behind your terminal does that. You provide the thin air of your imagination. AI term itself is a bit of a misnomer - it's more like machine learning on steroid. Of course, you have to feed the machine to learn on.

:Niranjan.

AI needs data to base its work on. The "guy behind your terminal" takes the program which was exposed to data by the developer, and therefore is not out of "thin air". The guy that developed AI programs in the 1980s.

 

[nmp]

AI needs data to base its work on. The "guy behind your terminal" takes the program which was exposed to data by the developer, and therefore is not out of "thin air". The guy that developed AI programs in the 1980s.

If as a lay person, if I wanted to use AI to create a picture using any of the programs available, do I have to feed any data or images other than defining what I want?

:Niranjan.

 

[Sirius Glass]

If as a lay person, if I wanted to use AI to create a picture using any of the programs available, do I have to feed any data or images other than defining what I want?

:Niranjan.

The programmer load the program with data to start it working, then after testing it, releases the program to the public. As each user added data, in this case scanning more art examples, the database grows. Therefore AI does not produce "out of thin air" or from a vacuum, but based on input data. The input data guides the results. It a different set of input data were used, say instead of Van Gogh, Rembrandt the results from the same program would yield wildly different results.

 

[nmp]

The programmer load the program with data to start it working, then after testing it, releases the program to the public. As each user added data, in this case scanning more art examples, the database grows. Therefore AI does not produce "out of thin air" or from a vacuum, but based on input data. The input data guides the results. It a different set of input data were used, say instead of Van Gogh, Rembrandt the results from the same program would yield wildly different results.

Nothing different to what I have been saying. I never said you don't need to input data. Never mind. Perhaps just semantic difference.

 

[Sirius Glass]

Nothing different to what I have been saying. I never said you don't need to input data. Never mind. Perhaps just semantic difference.

One CAN input data. The developer has started it by inputting data.

 

[DREW WILEY]

There is a "museum" full of Kincade paintings. It's a ridiculously Pepto Bismol pink-paint Victorian building in Pacific Grove he once owned. You'll never catch me in there. Even the outside color of the building repels me.

 

[MurrayMinchin]

Aaron Spelling also got rich by tapping into the lowest common denominator...I'd give an AI sitcom a try before watching one of his re-runs.

Is bad AI art as bad as bad human art?

 

[cliveh]

Some folk, like the OP, feel that AI is cheating.

I didn't say that, as I just regard it with a different value to that produced directly by human beings.

 

[snusmumriken]

There is a "museum" full of Kincade paintings. It's a ridiculously pink-paint Victorian building in Pacific Grove he once owned. You'll never catch me in there. Even the outside color of the building repels me.

I just had to google Kincade, but now I completely understand why you’d avoid any further exposure.