I want all the money I put in to be refunded in nonsequential bills, and give it to me in one of those fancy metal briefcases please.
I want all the money I put in to be refunded in nonsequential bills, and give it to me in one of those fancy metal briefcases please.
Weird how many here calls this Ai with no real proof.
The only current evidence I see that this might be Ai generated is blur lines around the character in the second and third panel, but the consistency of such a stylized drawing remaining the same is so strong it’s either really well done Ai that included photoshopping or human art with photoshopping.
A lot of you saying it’s AI clearly don’t work with it or have any understanding of it, because getting this level of consistency across 3 panels is very difficult. We’re talking about doing extra training involved around this specific character just to make sure you can get this type of consistency with 3 separate poses (my experience with it so far is in photo editing along with other digital editing tools).
I feel like recently I’ve noticed a lot of people witch hunting real artists who make a small mistake or have a unique art style (like that of the webcomic There Are No Demons), and the default of calling anything generated by AI, even something that was actually well done, “slop” shows less of an actual analysis of the piece but blind band wagoning
this specific style has been proliferating the past few weeks. there’s some new model that’s really good at just two-tone comics. it’s immediately obvious when doing a side-by-side.
Reverse image searching this reveals this comic was first posted on facebook four months ago, and a mix of similar AI and human drawn comics, so yeah I would say that checks out.
This is better evidence, but still would be difficult to keep such consistency with the character. That would be quite the advancement itself.
It’s absolutely AI. The inconsistent arm length, the distorted/melty flowers in the third panel, the bottom of the TV stand, some other kinda melty/bendy lines here and there. The only thing to discuss really is whether or not pure AI art that isn’t sloppy like this comic should be allowed or not because of the ethical implications of AI.
Because it is impossible for humans to get anatomy wrong, or just not care. Right panel is a compromise: Gesture wants negative space to the right (gal is running), panel width is limited, don’t want to cut off hand, so make arms different length it’s a comic FFS.
Because everything humans paint is hyperrealistic and infinite detail. Impressionist technique doesn’t exist.
…what? Best guess I have is that you think that it’s geometrically wrong but it just doesn’t have a door, drawers, or such. Which btw would break the composition.
Smudge tool don’t exist.
Can you not see the image I posted in the comment? Anyways, the flowers are not merely blurred or smudged. The way the petals are drawn is not how a human would draw them. Looks at the lines on the petals. Don’t think that I mistake imperfection for AI generation.
Yes and I can’t make heads or tails of what you want to say with it.
https://www.youtube.com/watch?v=jgDYlmRwXF8
What I’m hearing from you is vague gesturing, “don’t think I’m confused about that kind of stuff” and “trust me, bro”.
“Not how humans would draw them” doesn’t cut it. Explain how it differs. Zooming in on those flowers “hah they don’t look like flowers” does not prove anything, of course they’re not perfect: They’re in the background. Too much detail would make them pop out, that’s why you draw the impression of detail, not actual detail. The artist in the video is a master at that.
Inconsistent arm length is only proof that someone might have issues with reference sizing, not AI. Webcomic Pixie and Brutus had it so bad the artist made it into a hilarious joke and then stated they’d finally make a size reference chart to keep it consistent in the future. Freaking animated series Steven Universe, a professional production, constantly had issues with not just limb lengths but size consistency as well. Size consistency is the worst reasoning of them all if anything.
The flowers in the background could just be that the artist has issues with distance perspective / drawing. A lot of artists who don’t do backgrounds have that issue.
And like I already mentioned, the best evidence so far would be the blur lines surrounding the character, which would be evidence of AI in that they added the character into these backgrounds, but not only could that also be just a standard blur tool (because the artist isn’t good at backgrounds), but it would be weird that someone using Ai to make this would go through all the work of training a custom LORA or model on this character specifically, but wouldn’t reduce the size of blurring in the parameters when inpainting. It’s possible, sure, but usually someone who already went to the lengths of learning how to train a model or such to achieve such consistent results would also quickly look up how to set the inpainting parameters to not show such a massive blur line.
But the other thing to note that this might not be AI is the aspect ratio. Presumably this would have been done as separate panels on AI, and then you’d crop if needed and make a fake comic panels. If so, why would you crop the edges of the comic / leave them incomplete? Usually that’s specifically where an artist of this type of comic panel would sign their work too, if they chose too. Cropping out signatures is something commonly done by people reposting work done by real artists, and has been an issue for years as well.
Now, all this is too say that it’s not definite that this is made by a human. The point is that there’s no strong evidence that it’s AI. The clothes are consistent, the shoes and fingers are as well. That there’s 5 fingers is actually suspicious, but it’s definitely something an amateur artist would choose to do.
To me, the most likely case is this is a mix of AI and real drawn work / lots of manual editing work (for example, the mouth in panel 2 and 3 look the same, indicating it may have been copy and pasted by hand rather than generated). But my second guess it’s purely drawn work with a lot of manual editing. What I can say is this is very unlikely to be 100% AI.
No human would draw those flowers, that is clearly AI
The facial expression in the first panel made me think of AI. It does not quite fit with the panel’s context. I would expect a surprised facial expression if not a down right happy one.
If you pay attention to her shirt, you can see diagonal folding lines on the clothes because presumably one of her shoulder is higher than the other. The artist made it consistent across three panels and it’s clearly a human’s intention.