this post was submitted on 06 Sep 2025
821 points (93.5% liked)
aww
25215 readers
439 users here now
A place with minimal rules for stuff that makes you go awww! Feel free to post pics, gifs, or videos of cats, dogs, babies, or anything cute and remember to be kind to others.
AI posts must be labeled [AI] in the title and are limited to one per week.
While posting and commenting in this community, you must abide by instance-wide rules: https://mastodon.world/about
- No racism or bigotry.
- Be civil: disagreements happen, but thatdoes not provide the right to personally insult others.
- No SPAM posting.
- No trolling of others.
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
I would say exactly the opposite - it proves the point. The sameness of the two dogs and the lack of the corresponding marriage ceremony in the background rob the image of most of its significance, and the background is a copy that wouldn't exist if the original hadn't existed.
This was more of a technical demo than anything else. The real issue I had was that the movie is still under copyright and I had a find a prompt that would allow it to make that much. But I wanted it be as identical to the original as possible to see if it could replicate the soul of the original.
It's not there I would agree, there is a certain... Flavor missing but it's very close.
More so I would agree that llms will not be able to ever replicate individual human nuance especially in art.
I really disagree. The lines and content are similar but the soul is completely lacking. This looks like it's been simplified so much so it could be a coloring book or something. The soul from the plants is totally gone , the coloring of the windows, I look at the generated version and understand completely what I'm looking at, but I feel nothing. P
Yes. Like I said art is extremely subjective, your personal a bias is playing a role. And it is lacking soul absolutely there's almost a metallic quality to it.
Once again I generated the picture for the sake of comparison. Technically it's very close to the original. More importantly because it's an llm it's missing nuance.
I don't use it but isn't that just additional prompts?
"Add a marriage ceremony shadowed in the background."
"Change the dog on the right."
I would think the entire style could be built up with enough prompts. It's no different than digital artists using photoshop and picking a brush that does "watercolor" and then applying a hand paint texture. I have made Photoshops that mimic Van Gogh with a few clicks.
I'm in the camp of if you aren't actually painting, you have no reason to complain about AI art.
Look closer. It's not just the big details. Look at the window. Look at how the AI just completely flattened out the lighting outside to this vague orange tone that robs the scene of so much of its character.
AI can't think. It can't make the kind of intentional choices that a good artist can. And you can't solve that with prompting, because one, good luck describing that exact pattern of lighting in a prompt, and two, the person prompting isn't a good enough artist to come up with that exact pattern of lighting, because it would take years of experience to be able to.
The reason you think your photoshops have mimic'd Van Gogh is because you're not Van Gogh, and no offense intended, but you're clearly not a good enough artist to understand the difference. I guarantee that anyone who knows what they're looking for would instantly know the difference between your work and his. Art isn't just style, it's a myriad of choices that you can't recreate with a photoshop filter.
No one claimed it does. It requires prompting from a human.
There is absolutely no difference between written commands and pointing a clicking from a menu on the screen. I see it like writing a program. If the code doesn't do exactly what you want, you need more code. No one today claims that a word processor can't do the same thing as a typewriter because a typewriter was assembled by hand as compared to a word processor that needed hundreds of thousands of lines of instructions to simulate the same effect.
I absolutely didn't mean to imply that it was. I was claiming that using Photoshop with all its artificial filters and brushes to simulate styles is the same as prompting an AI to simulate those same styles.
And those choices can be written as text or picked from a menu.
The part you're still not grasping, despite my already having explained it once, is that in order to create this perfect prompt you're imagining - in order to actually be the person who would make all those choices - you'd have to have the kind of experience that only comes from years or decades of practice. Is there some version of AI art where experienced artists use it as their medium in lieu of pencil or oil paint or digital art? Maybe. But that's not the point of all this, is it? The promise of AI is that it's supposed to allow everyone and anyone to be Van Gogh, without any training or practice, but the person who has no training or practice is never going to be able to create that perfect prompt that you somehow imagine exists.
(All of which is putting aside that when you move a brush the paint goes where you want and is the colour you want, whereas a prompt will always be filtered through the random distortion field of a stastical association model, but we don't even need to get into that)
You are misreading what I have written. I even explicitly said "No one is saying that." In response to your earlier claim.
Yes. Exactly.
That's not the argument I made! You are way into strawman territory.
I claim that AI can be art because it is a tool that requires human intent. The type of effort is irrelevant. Everyone against AI art dismisses the effort of writing careful exacting instructions to achieve a result. Movie directors are considered artists despite not sewing the costumes, painting the sets, or writing the words. They only prompt others into achieving their vision.
How many rough drafts of 101 Dalmations would you watch before arriving at what is now the final product?
It isn't just the watercolor. It also is the decision by the artist to use watercolor.
The AI and its user did not know to add those 'additional prompts', and for that the result suffered. The result is then discarded for the next iteration which defines it as: slop.
Agreed.
If effort is art than again digital artists aren't real artists because they aren't using oil on canvas.
Painters that don't mix their own black aren't real artists. Inscribing words without deference to God is to blaspheme.
Art is not necessarily effort, but effort can be art. While I am agreeable that AI is a tool, it is one that appears to be used quite ineffectually. I'd love for artists to be able to produce their own stylized models. That'd be the legitimate pathway genAI might be able to take to follow the same trajectory as photoshop did for digital artistry.
Nonetheless a digital artist can convey their effort in a manner of the same order as oil on canvas. The problem with genAI is that it is purely derivative. It is limiting to the creative process as a result until the models, checkpoints, and such are entirely rooted into it.
If you didn't make the model, train the checkpoints, and refine the prompts then you're just feeding a black hole of derivative work, every time.
Honestly the hallucinations and horrors are the only things genAI have been creative about.
I’m choosing to believe you’re just a troll because it takes real human effort to draw both digitally and traditionally. I can tell you’re not an artist and have never spent hours—years even—practicing.
Not to mention, the fact that all AI image generators are using stolen art, made by real humans, who wanted to share their ideas, their style, their added experiences, their perspectives, is why the image generated looks soulless, bland, and drawn with no real substance to many who can tell when something is AI gen.