So her pupils are smushed together, there’s a weird color near her collar necklace along her arm in the second panel, like if it was going to be skin but was changed to be part of the sleeve last minute. The Bird has a silver piece between it’s eye but it kind of turns into a badly drawn eye in the second panel, comes back on the third but it’s off and then turns into a misplaced thicker red strip on the forth.
These are the exact types of errors that an AI would make or used to at least. Cross eyed look and parts that seem to sit in between two things like if it wasn’t certain if it should do one or the other.
I get he makes everything himself, but coupled with the style, it did really look generated the first time I saw it.
I’ve found that focusing on really small details like that is kind of useless , especially in mediums like comics where it’s common to be working on a deadline; If you are start applying that logic to the frames of animation done on a budget you are going to conclude that every animation is AI, because when an artist is working fast and loose they tend to leave artifacts like those behind. On Revoy’s case; His style is much too consistent and his palettes tend to be very muted compared to most AI comics out there. I don’t read his comics much but it is very clear to me when he is doing a new “arc” usually because he keeps the color palette of his comic arcs very consistent. I think he color keys his arcs or something.
I’m not one to go around flagging people as AI, or harassing artists; But I found that the biggest tell for people relying on AI Is a lack of aesthetic direction; and a lack of unique “manerisms” or “short-hands”. Mannerisms was a term used for Italian painters that has very persistent stylization patterns, People like “el greco” that had this tendency for elongated figures and sharp facial structures. Most artists have mannerisms; Even when trying to copy someone else’s style and the more advanced an artist is, the more their mannerisms show. It’s like hand writing.
Short hand is a term you will see a lot in animation. They are like quick personal recipes to imply complex structure with ease. A certain way to draw a hand, or a certain way to convey the eyes in some expressions. The more advanced an artist is; You start to find their short-hands again and again because they have become part of their drawing process.
So the issue With AI is that, when looking at a bulk of work; You can’t find these things; Even when loras are heavily referencing other artists. Ai image models do this really weird thing where they average a bunch of mannerisms and short hands that are becoming signature to diffusion models themselves. Something about the way they seem to convey certain expressions again and again; A certain AI surprised and rage face, a certain "cute AI smile that always comes up in “adorable characters” It’s almost like the model picks up in a series of mannerisms itself and it is very hard to go around them. I’ve seen people that use LORAs or specialized training to circumvent this and in the end some pieces look like the bias of the original model and some others look like the bias of the LORA.
So basically look at the bulk of work from an artist; Even when they change mediums or style there is always this human uniformity to them; These certain limitations in their chosen aesthetics that are hard to pass by.
I don’t Really follow Revoy’s work because It’s a bit too cutesy for my personal taste, but I can identify one of his comics on the middle of a doom scroll just by his mannerisms. AI image generations tend to have this thing where they bleed to each other, It’s either hard to sort them out from the base midjourney/stable difussion style; Or they borrow so much from specialized training that you see the artist of origin. Then again; I’ve been an art fan for years. So i might be particularly sensitive to this.
In the end, trolls that want to legitimize AI will attempt to blur the lines between synthetic generation and human artists. They will attempt to steal the styles from the most successful artists and then accuse them of being in their bandwagon. It is a very scummy tactic. The people that rely on AI need either to be legitimized as “artists” or to bring down everyone into the mud with them; So i would treat people who accuse others of AI usage especially as suspicious; Check their post history; Look for their galleries, See how connected they are to the artist community. If they are a new rando that follows no one, has never commissioned or done work themselves and this person is doing the rounds raising controversy, You might have found a troll account!
So her pupils are smushed together, there’s a weird color near her collar necklace along her arm in the second panel, like if it was going to be skin but was changed to be part of the sleeve last minute. The Bird has a silver piece between it’s eye but it kind of turns into a badly drawn eye in the second panel, comes back on the third but it’s off and then turns into a misplaced thicker red strip on the forth.
These are the exact types of errors that an AI would make or used to at least. Cross eyed look and parts that seem to sit in between two things like if it wasn’t certain if it should do one or the other.
I get he makes everything himself, but coupled with the style, it did really look generated the first time I saw it.
I’ve found that focusing on really small details like that is kind of useless , especially in mediums like comics where it’s common to be working on a deadline; If you are start applying that logic to the frames of animation done on a budget you are going to conclude that every animation is AI, because when an artist is working fast and loose they tend to leave artifacts like those behind. On Revoy’s case; His style is much too consistent and his palettes tend to be very muted compared to most AI comics out there. I don’t read his comics much but it is very clear to me when he is doing a new “arc” usually because he keeps the color palette of his comic arcs very consistent. I think he color keys his arcs or something.
I’m not one to go around flagging people as AI, or harassing artists; But I found that the biggest tell for people relying on AI Is a lack of aesthetic direction; and a lack of unique “manerisms” or “short-hands”. Mannerisms was a term used for Italian painters that has very persistent stylization patterns, People like “el greco” that had this tendency for elongated figures and sharp facial structures. Most artists have mannerisms; Even when trying to copy someone else’s style and the more advanced an artist is, the more their mannerisms show. It’s like hand writing.
Short hand is a term you will see a lot in animation. They are like quick personal recipes to imply complex structure with ease. A certain way to draw a hand, or a certain way to convey the eyes in some expressions. The more advanced an artist is; You start to find their short-hands again and again because they have become part of their drawing process.
So the issue With AI is that, when looking at a bulk of work; You can’t find these things; Even when loras are heavily referencing other artists. Ai image models do this really weird thing where they average a bunch of mannerisms and short hands that are becoming signature to diffusion models themselves. Something about the way they seem to convey certain expressions again and again; A certain AI surprised and rage face, a certain "cute AI smile that always comes up in “adorable characters” It’s almost like the model picks up in a series of mannerisms itself and it is very hard to go around them. I’ve seen people that use LORAs or specialized training to circumvent this and in the end some pieces look like the bias of the original model and some others look like the bias of the LORA.
So basically look at the bulk of work from an artist; Even when they change mediums or style there is always this human uniformity to them; These certain limitations in their chosen aesthetics that are hard to pass by.
I don’t Really follow Revoy’s work because It’s a bit too cutesy for my personal taste, but I can identify one of his comics on the middle of a doom scroll just by his mannerisms. AI image generations tend to have this thing where they bleed to each other, It’s either hard to sort them out from the base midjourney/stable difussion style; Or they borrow so much from specialized training that you see the artist of origin. Then again; I’ve been an art fan for years. So i might be particularly sensitive to this.
In the end, trolls that want to legitimize AI will attempt to blur the lines between synthetic generation and human artists. They will attempt to steal the styles from the most successful artists and then accuse them of being in their bandwagon. It is a very scummy tactic. The people that rely on AI need either to be legitimized as “artists” or to bring down everyone into the mud with them; So i would treat people who accuse others of AI usage especially as suspicious; Check their post history; Look for their galleries, See how connected they are to the artist community. If they are a new rando that follows no one, has never commissioned or done work themselves and this person is doing the rounds raising controversy, You might have found a troll account!