Hacker News Reader

Comment by StewardMcOy | Hacker News Reader

I know this article is very informative, and I typically like to keep discussion substantive, but that artwork at the top really distracted me from actually reading for a good few minutes.

I don't really care of it's AI or not. The picture is just strange. The empty hoodie just standing by the window, watching a worker who is seemingly undisturbed by it. The odd placement of the desks, the fact that all of the text is gibberish, but looks just close enough to real that you pause to examine it. Except for the big "CHECK" on the wall, which isn't something you'd actually see in a real establishment that cashes checks.

More and more, I see companies using art that they would refuse to pay for if a contracted illustrator turned it in. Seemingly because it's AI, it's like they become blind to how bad it is.

View on HN

Yikes, it really is distracting. What really gets me is the screwed up, subtly M C Escher like perspective.

Also, yeah, this looks like a weird bank / accountant's tax office pastiche, not at all like a check cashing place.

It's patio11 though, all of whose writing on the nuts and bolts of finance are fascinating, so it's easily forgiven, heh

Edit: in fact I'm already 3 paragraphs in and so fascinated by the, typical for patio11, high level of detail combined with easy novel like readability, that I almost feel guilty for having piled on to criticize the awful AI image. But still, yeah, I agree AI images like this are actually worse than no illustration at all

Right? That's why I was hesitant to bring this up too. But the picture really is that distracting.

Like, the article's good enough that it doesn't need an image. Or if you really think you need it, grab some free stock photo or clipart or something.

Or get good AI art. Don't settle for bad AI art.

Yeah, at least touch up the text so it isn't just gibberish (besides for "CHECK"). But I guess this is what happens when someone with (presumably) no graphics skills gets an image out of AI - they are powerless to tweak it.

This is a blog post by a rather small operation (probably one guy). If AI-generated-imagery weren't available, he would've used a stock photo or nothing at all. I doubt very much he would've paid an illustrator for it. So no he's not blind to it, he just doesn't care, or has decided that the low quality is worth the low price. Analogous to the situation in the article.

This is a perfect example of what I’ve been saying for a few years: Artists’ value is almost entirely in their ability to figure out exactly what should be in a piece to visually communicate an idea— not in their ability to make the pieces themselves. And it’s not even like this ill-conceived generated graphic is better than no graphic at all… Even in a blog post, the author’s lack of visual vocabulary makes the whole piece worse. Is not just AI— you can certainly buy the wrong stock photo, for example, but the photographer’s skill and the service’s curation makes mini golf out of a golf-sized task.

>Artists’ value is almost entirely in their ability to figure out exactly what should be in a piece to visually communicate an idea— not in their ability to make the pieces themselves.

Practice has shown the two to be inseparable, presuming you mean artistic value rather than market value. They inform each other. Furthermore, no valuable idea or piece of art can be owed to a single person.

You've missed the point, entirely. I'm not talking about credit, I'm talking about capability. By any measure, the image at the top of the article is polished and it grabs your attention, and if an artist had made it by hand, even experienced artists would commend their solid technical art skills. It also sucks so much at accomplishing its job as a communication tool that it detracts from the article content for many people. A layman having powerful image generation tools doesn't make them capable of communicating ideas visually any more than copilot makes laymen capable of designing sane software architecture, or having Grammarly would make them any better technical writers. People that don't think so don't realize how much they don't understand about solving communication problems with imagery. There are plenty of lifelong fine artists that could whip my ass making compelling imagery in physical media, but would still need years to learn the fundamentals of visual communication. It's a set of skills that need to be learned and practiced, and making the images themselves is probably the least consequential part of it. Art directors generally don't make any images at all, and the idea that someone could step in and do their job because they could whip up a bunch of slick-looking junk at a prompt is beyond hilarious.

What is an "art director" though? We're talking about artists. There's a reason they don't share the same title and everybody knows it.

Auteur theory exists for good reason. Good art cannot be imagined merely in the mind, only self-flatulating bullshit can. One must engage and prove and iterate and exercise all the unlikely leaps that must be made to go from medium to meaning to detail. Image generators are just shitty art directors and neither is practicing art.

How glib. In the future, when avoiding addressing everything someone says, you can save yourself some time by not entering the conversation thread.

The image is worth a few comments. No artist would ever do this. The text, the stool legs, the paperweights, the wristwatches, the shoes, and all other little details are quite terrible.

Bloggers will swallow a lot for four cents (the cost of a single decent Hd DallE3 image generation via the API).

Actually I think they edited in the "CHECK" sign, the other texts in the image (the sign of the shop across the street, the text on the terminal to the left) are the usual "AI-generated gibberish".

Reply on news.ycombinator.com