If you simplify it down to: "AI only copies this style 1 to 1" then you are correct. But AI is trained on many things and throws it all together like fruit salad. Which means.. again.. that you, the loud mouth, are wrong, and my analysis stands.
I can see that things will have to be spelled out in excruciating detail because thinking for oneself is not a skill you have done much to work on. Lets go through why your arguments are complete trash step by step, and I'll quote your idiotic premises so you can't just fall back on "if you summarize everything I said in a clear and concise manner it makes me sound dumb so that's not what I mean!" (these are sarcasm quotes, I am not directly quoting you here, I don't want you to be confused).
Well, kudos to you offering a look behind the curtain. But, I'm afraid, it didn't manage to dispell my suspicion entirely. I will provide why.
Well, at least we know you saw the post, but lets see if you processed any of the info there...
For that I ask you to just look up any female.. picture made with generated AI. Focus on those details:
Eyes - shape, volume
Cheeks - sharpness, colouration
Nose - Shape, colouration
Mouth - Form, shape of the smile, lips, colour
Chin - pointyness, jawline
As we can see, no, you did not address a single thing presented in the info about why this was not AI. There are criticisms to be found there, and I'll get to it eventually, but not from you. Instead, you've focused on what is just generic anime style. This is a style that has been true of every generic anime styled game for 20 years. It's a style so prolific that all of the basic art generators out there mimic it because, again, they were trained on it. None of this is evidence of AI in and of itself. It's enough to make one wonder if it might be until evidence is provided one way or another, but not nearly enough to draw any conclusions from.
You will find many generated pictures that have many of the elements you showcase with your new tinkerbell. Big eyes, bright coloured iris. Light red, smooth cheeks with freckles. Small stubby nose, light red at the tip, small to no bridge. Reddish, smooth lips. Sharp clear jawline, small pointy chin.
This might be coincidence. I give you that. But it would be a rather odd one. This could also mean that you edit the generated picture heavily, with just the examplatory template being generated.If you and your art team are truely innocent, I am sorry. But that well is utterly poisoned, sadly.
Good luck with your project.
I cannot fathom why this would be an odd coincidence. Art based on the same anime style you can find across the internet for the last 2 decades looks similar? WHAT ARE THE ODDS?! I guess 99% is too small a chance to be coincidence.
"Edit the generated picture heavily" And you're accusing me of not knowing how AI works. Dear crap.
I've installed stable diffusion and multiple different loras and messed with AI for fun enough, but I have lots of other exposure to it. I am in no way an expert, but after reading this I feel like one in comparison.
It is entirely possible that a reference images was generated and then used to draw their own version, or even traced... but reference imagery is basically true of any art out there, and I don't think that is particularly a problem (maybe the tracing scenario to some extent.)
Now if you want to properly criticize this as potentially AI, you can throw out all of the very not smart things you listed and actually look for real signs of AI. For example, it's real suspect that none of the drawings of the face in the evidence post have any sort of sketch or structural drawings to view... it's just pasted onto other sketches. If you look at the detail of some of the images posted, the linework has varying levels of quality... some are super fine and others are thick and bulky in ways that I don't think make sense, which is strange. There aren't the usual issues where eyes don't match and obviously we don't have any fingers to count. The positions of the lines make sense, which when they do not is another sign of AI. AI doesn't *know* what it's doing, it doesn't *know* all the lines need to connect, especially internally in am image, it's just using heuristic models, so you get eyes that are different, lines that don't quite connect like they should. It gets pretty easy to spot with practice and exposure to lots of them.
What the evidence post does have, that you did not address and why your follow up arguments are as terrible as they are, is the image editing software that has broken the face into tons of layers. AI doesn't do that, AI spits out a full image, there is some nuance there where you can do lines and then coloring, but you're not going to get the different shade layers and hair lock layers they have separated to a degree I find almost comical. You have to do that on your own, and if you can do that then tracing the AI is basically more work than just draw and coloring on your own.
AI is an awful tool for consistency, so one image will make you question but once the game comes out you'll have multiple images of the same character from different angles and perspectives and you'll be able to tell pretty quickly for certain whether someone is using AI to draw this stuff as the details just won't match, you can see that with any AI game on this site.
