AI will turn into a tool, like any other. Like tools, the results will vary with the craftsman. Similarly, tools can be used for bad as well as for good.
Am I the only person old enough (I'm 65) to remember the hair on fire screaming about computer assisted art and computer drawing programs?
If you go further back, photography came along. I believe that the visual arts are STILL trying to adjust to the fact that your grandmother can take a good portrait of someone using a camera instead of spending a decade learning how to draw well.
gAI still has a lot of problems with words. There are a couple of specialized models that can do good/decent words. you are also asking it to change the style, so it's making more changes in line with that style. To get things more similar, maybe ask for Royo, or something like that and remove all the words so it doesn't confuse them as other articles in the picture.
More descriptions, tweaking and in-painting (like Adobe Firefly does) will get it closer.
A pay for hire artist would do the lettering separately, so to make a fair comparison you would need to have a "clean" picture to start with.
That said, gAI NEVER comes up with something better than what I have in my head, or even WHAT I have in my head. But if someone doesn't have an artistic eye, or they haven't done much/any drawing themselves, then, yeah, it seems like magic and oh so pretty! And they pretty much all look the same.
That's why I never use gAI in my cover art designs. It's HMA (Human Made Art).
Pretty much my take whenever I play with AI art. It's a really fun toy, and if used carefully, it can be a useful tool. But in one of my AI Programming courses I took many years ago, my professor said something that is as true today as it was way back when. "Creating artificial intelligence is easy. Preventing artificial stupidity is the hard part."
I think some of the outputs speak for what the general inputs and accepted outputs have been. Especially these public facing sites, that take a lot of input. It's producing for the average, based on the input material. Not the straight average... there is some variability in the above... but generally the average. If you vaguely know what that is, you can detect the uncanniness even in what are quite good images, but have the subtle touch of the machine.
In the specific case of your covers... there's a lot going on. The letters and their order contain specific meaning, the superposition of stuff against the background, various other items... past a certain point the machine really falters, because it doesn't understand the different layers of stuff. It doesn't know it should copy letter for letter. It doesn't grok the contrasts of styles. I notice the foreground tends to be all generation really cares about, and the background tends to be nonsensical stuff. And since the machine doesn't have that seperate layer mindset... you can't tell it "keep this background, reroll the foreground."... you can try, I just don't think it'll work.
I sat and laughed at these. I can't get AI to do what I want, either. I use Bing AI and Grok because they're free, and Grok doesn't have as many weird restrictions as Bing. (Bing will absolutely not generate a person riding a motorcycle, apparently that term is unsafe). Also I thought it was chatgpt that had been trained specifically on Ghibli? I ought to try this kind of post with my own book covers. :-D
I use AI comic-book style for my characters because it's more forgiving, particularly for my non-human characters. It's also a great reference for when I describe a character. Also, when I hire an actual artist, I can show him the pictures. "This is what so-and-so looks like except his hair is a lighter shade and he has multiple facial piercings. In the picture, he is brandishing two laser pistols." If used correctly, it's a great tool.
It makes the artist's job easier, IMO. With the AI “sample” in front of them, they don't have to guess what the author wants. The way I see it, it saves time they could use for other projects.
AI art programs are only as good as the inputs. If you can't get them just right, good luck.
It's good for pictures, I guess. Oh well.
AI will turn into a tool, like any other. Like tools, the results will vary with the craftsman. Similarly, tools can be used for bad as well as for good.
Am I the only person old enough (I'm 65) to remember the hair on fire screaming about computer assisted art and computer drawing programs?
If you go further back, photography came along. I believe that the visual arts are STILL trying to adjust to the fact that your grandmother can take a good portrait of someone using a camera instead of spending a decade learning how to draw well.
People adjust.
gAI still has a lot of problems with words. There are a couple of specialized models that can do good/decent words. you are also asking it to change the style, so it's making more changes in line with that style. To get things more similar, maybe ask for Royo, or something like that and remove all the words so it doesn't confuse them as other articles in the picture.
More descriptions, tweaking and in-painting (like Adobe Firefly does) will get it closer.
A pay for hire artist would do the lettering separately, so to make a fair comparison you would need to have a "clean" picture to start with.
That said, gAI NEVER comes up with something better than what I have in my head, or even WHAT I have in my head. But if someone doesn't have an artistic eye, or they haven't done much/any drawing themselves, then, yeah, it seems like magic and oh so pretty! And they pretty much all look the same.
That's why I never use gAI in my cover art designs. It's HMA (Human Made Art).
Pretty much my take whenever I play with AI art. It's a really fun toy, and if used carefully, it can be a useful tool. But in one of my AI Programming courses I took many years ago, my professor said something that is as true today as it was way back when. "Creating artificial intelligence is easy. Preventing artificial stupidity is the hard part."
I think some of the outputs speak for what the general inputs and accepted outputs have been. Especially these public facing sites, that take a lot of input. It's producing for the average, based on the input material. Not the straight average... there is some variability in the above... but generally the average. If you vaguely know what that is, you can detect the uncanniness even in what are quite good images, but have the subtle touch of the machine.
In the specific case of your covers... there's a lot going on. The letters and their order contain specific meaning, the superposition of stuff against the background, various other items... past a certain point the machine really falters, because it doesn't understand the different layers of stuff. It doesn't know it should copy letter for letter. It doesn't grok the contrasts of styles. I notice the foreground tends to be all generation really cares about, and the background tends to be nonsensical stuff. And since the machine doesn't have that seperate layer mindset... you can't tell it "keep this background, reroll the foreground."... you can try, I just don't think it'll work.
I admit I laughed more than I probably should comparing the pictures!
I sat and laughed at these. I can't get AI to do what I want, either. I use Bing AI and Grok because they're free, and Grok doesn't have as many weird restrictions as Bing. (Bing will absolutely not generate a person riding a motorcycle, apparently that term is unsafe). Also I thought it was chatgpt that had been trained specifically on Ghibli? I ought to try this kind of post with my own book covers. :-D
I use AI comic-book style for my characters because it's more forgiving, particularly for my non-human characters. It's also a great reference for when I describe a character. Also, when I hire an actual artist, I can show him the pictures. "This is what so-and-so looks like except his hair is a lighter shade and he has multiple facial piercings. In the picture, he is brandishing two laser pistols." If used correctly, it's a great tool.
And some artists, like GPrime (whose last name is long and Greek) will tell you to screw off if you do that...
Which makes no sense, since I've done that with my wife, who did the Neck Romancer cover.
(Shrug) Artists
It makes the artist's job easier, IMO. With the AI “sample” in front of them, they don't have to guess what the author wants. The way I see it, it saves time they could use for other projects.
I concur