Hacker News Reader

Comment by ARandumGuy | Hacker News Reader

I feel like the comment made by the author's friend captures a lot of my feelings on AI art. AI art is often extremely detailed, but that detail is often just random noise. It's art that becomes worse and makes less sense the more carefully you look at it.

Compared that to human art. When a person makes a highly detailed artwork, that detail rewards looking closely. That detail forms a cohesive, intentional vision, and incentivizes the viewer to spend the time and effort to take it all in. A person could spend hours looking at Bruegel's Tower of Babel paintings, or Bosch's "The Garden of Earthly Delights".

Overall, I've never felt the need to spend the time looking closely at AI art, even AI art that I couldn't tell was AI right away. Instead of rewarding close inspection with more detail, AI art punishes the viewer who looks closer by giving them undecipherable mush.

View on HN

The gate picture has the same problem as as the cat one that he didn't filter out. There's a lot going on, and the lighting does seem to be one of the somewhat-inconsistent issues IMO, but it's just generally weird about why there's cats of all different sizes, why some of the smallest cats have the same coloring as the biggest ones, but some don't, what's going on with the arms of the two darker cats on the right, why aren't the sides of the throne symmetric, etc.

Everything is consistent in terms of "the immediate pixels surrounding it" but the picture as a whole is just "throw a LOT at the wall."

It passes the "looks cool" test but fails the "how likely would a human be to paint that particular composition" test.

The cat picture really shows the "noisy detail" problem with AI art. There's a lot going on on the area directly above the cat with a crown, as well as on the armrests and the upper areas of the wall background. But it's all random noise, which makes it both exhausting and distracting. A human artist would probably either make those areas less detailed, or give a more consistent pattern. Both would let those parts fade into the background, which would help draw our focus to the cats and the person.

There's other, more general issues too. The front paw on the big cat on the left is twisting unnaturally. The cat on the right with the pendant thing looks like it only has one front paw. The throne looks more like a canopy bed then a throne, with the curtains and the weird top area. The woman's face is oddly de-emphasized, despite being near the center of the piece.

Most of these things are subtle, and can be hard to articulate if you aren't looking closely. But the picture reeks of AI art, and it doesn't surprise me that the author was able to identify it as such right away.

Reading this comment chain is kinda confusing to me. Y'all really not aware that there are countless works of art that predate AI art that look literally just the same?

Missing symmetry is super frequent in hand drawn art, and overusing/adding too much detail that ultimately detract from the art is something any aspiring artist has done at some point.

I get that you don't like to look at pictures like that, but it's really not unique to AI art.

My point wasn't that you can't find art made by a human that has the characteristics of AI art. My point was more that it's just bad art.

Asking LLM for thought processes only generate hallucinations. Spotting AI images are same. Those subtleties are justifications ex post facto, not necessarily the actual cues that trip BS detectors.

Maybe the author's friend is just way better than me at this, but I tried applying her advice to some of the other images and I don't feel like it would have helped me.

Looking at the human impressionist painting "Entrance to the Village of Osny" that lots of people (including me) thought was AI, it seems like there's lots of details which don't really make sense. The road seems to seamlessly become the wall of a house on the right side for instance. On the other hand, even looking at the details, there's no details in the cherub image that I could see which would give anything away.

It’s impressionist. It’s not supposed to make sense in the sense that it’s an accurate reflection of reality; it’s supposed to make sense in that you can understand why the details were drawn in the way they were because someone put thought and intention into them.

I guess a good example is things like the physical characteristics of a line drawn by the stroke of a paint brush. You can often see they all align generally in the same direction, and have a sort of "fingerprint"

I was able to tell because the distant houses are placed in a nonsensical formation in the AI image, but in the human image they make sense (they're more of a 'swoosh').

The arms look, er, not very feminine.

This is why I'm baffled when people want to put this kind of stuff on their Behance/Artstation profile.

Can AI art be useful? Sure but I'd argue only in the pursuit of something else (like a cute image to help illustrate a blog article), and certainly not for art's sake.

Posing it as "ART" means that the intent is for viewers to linger upon the piece, and the vast majority of AI art just wilts under scrutiny like that.

It's exactly what isn't captured in the training data. The AI knows what the final texture of an oil painting looks like but it doesn't, know if what it's creating isn't possible from the point of view of physical technique. Or, likewise, it doesn't see the translation from mental image to representation of that image that a human has. It's always working off the final product.

That makes it sound like impressionism. But the phony details have a more intense bullshitting quality, like the greebles on a Star Wars spaceship.

There's a lot of thought that goes into things like the greebles on a spaceship, like the shape language, the values and hues, etc.

Impressionism might seem "random" like what a model would output, but the main difference is the human deciding how that "randomness" should look.

The details on a model generated art piece are meaningless to me, no one sat down and thought "the speckles here will have to be this value to ensure they don't distract from the rest of the piece."

That's more what I look at when I digest art, the small, intentional design choices of the person making it.

Hmm? Impressionism is noted for extreme lack of detail, that still is suggestive of something specific, because the artist knows what details your brain will fill in. (8bit pixel art is impressionistic :-) )

Yes, a suggestive smudge, a vague mark, as the artist in the article said (the one talking about the "ruined gate" picture). That's like an honest communication between artist and viewer, "this mark stands for something beyond the limit of my chosen resolution". It's like a deliberately non-committal expression, like saying "I don't know exactly, kinda one of these". In contrast, we have in AI art misleading details that contain a sort of confabulated visual nonsense, like word salad, except graphical. Similar to an LLM's aversion to admitting "I don't know".

Reply on news.ycombinator.com