(D)Readful Things
Adam Perocchi & the trappings of AI generated art
Adam Perocchi AKA 'Readful Things' is an artist who makes a bunch of very cool custom pop-culture figures, posters, paintings etc. And they're good! Unique, fun ideas, with a load of thought, time & skill.
You can check out his work on Instagram, Twitter or his website.
Adam has been making cool stuff for a while now and, as revealed in a recent podcast, it’s his full time job. He sells these things every week on eBay and makes enough money to do nothing else. Almost every single artist’s dream!
Not only does Adam come up with the brilliant concepts for his ‘pieces’, he also designs them, sculpts, prints and paints the figures and, most impressively of all, creates bespoke, hand painted (sometimes digital, sometimes traditional) original artwork for them all. His artwork is so good that he even sells prints and posters!
It seems crazy that just one man can master so many disciplines. So how does he do it? It’s simple, he doesn’t.
Adam has been using AI to ‘help’ generate the majority of his artwork for the last 18 months or so. For anyone familiar with the telltale signs of AI generated imagery, this has been very obvious. Especially when you consider how many of these things Adam is churning out. An auction a week doesn’t give a lot of time for putting on a smock or sweating over a Wacom tablet, so Adam’s been turning to generative AI to get the work done.
So what?
Yeah, so what? It’s not like Adam hides the fact that he’s doing this. Except, he does! Every single time Adam is confronted about the use of AI in his work, he doubles or triples down, gets angry and liberally blocks those who dare to point out they can see the Emperor’s sack.
He also regularly manipulates imagery to help sell the illusion that he is painting this stuff BUT I’ll get to that later…
Proof
There is a LOT of proof but for simplicity’s sake, let’s start with this awesome poster concept Adam made
To the uninitiated, this looks like a cool shot from a VERY cool movie. Except, if you’re going to paint this or even Photoshop it, there’s no way you would accidentally paint two completely different trucks in each lens.
Look at them! At a cursory glance they’re fine but they have different cabs, grills, lights… No human would do this and honestly, it’s a little sad that Adam didn’t even spot such an obvious mistake.
But this is just ONE image and it’s not proof that he uses AI to the extent that I’m suggesting. Heck, I’m not even sure if it’s still online.
So here’s a buttload more!
HANDY
As we’ve established, Adam is a master of 3D sculpting, painting figures and portraits and generally working in ANY medium he can lay his talented hands on.
Speaking of hands, how many fingers ARE there on a hand again? Columbo’s pointing hand here seems to suggest SIX…?
Hey, this is a mistake any of us could have made. Maybe Adam just struggles with pointing at stuff?
SO Adam just loses count of fingers and then deletes the images when he realises but you know what else keeps adding fingers to hands? If you said “Generative AI models!” then congratulations! You can have a cookie. In fact, you can have SIX. One for each finger…
Moving on to this image Adam shared to Threads. Donald Pleasence was an incredible actor, but even a man with his range would struggle to make his hands work this way…
This is of course, a twisted version to Alex Ross’s own homage. Comparing the two, you can see that Adam’s quick AI fudging has messed up superman’s belt and boots
There are also SO many examples of Adam making decisions that very well COULD be his own but are so WEIRD that it’s hard to imagine anyone with his apparent skill-set/draftsmanship making them consciously..
No matter what your skill level, there aren’t many people who would draw this skeleton’s arm bones connecting this way.
Ask yourself, you YOU add a little extra tube that plugs into a hole…? WEIRD.
MOOOOORE
This is a cool Halloween III poster that Adam made. It looks great! Yet it’s almost certainly AI generated.
The snake branching into sections and phasing in and out of the pumpkin. The scale patterns make no cohesive sense (something AI still struggles with). The snake being weirdly cut off as it enters the eye. The screen right top side of the pumpkin is in a perspective that doesn’t make any sense.
These are honestly WEIRD mistakes to make but especially if you have the skill to pull the rest of it off. If you’re rendering light and shadows to the level Adam is here, you would notice the other problems. He didn’t, because AI did it for him.
The polish of this version hides a lot of the problems BUT I found another pass he made at it. This early version is screaming AI at us and has all of the above issues dialed up to 11.
And to be clear, this poster is a commercial project, made under license and sold to the public via Bottleneck Gallery NYC. People spent their own actual money on these, as they do with a LOT of Adam’s work!
BUT if Adam made this with AI (which I am very confident he did) then no-one can own the copyright to it. Remove the branding and anyone can sell this. Did he sign a legal document with the license holders stating it would all be his own original work, because I’m pretty sure he would have!
Another side to this is that Bottleneck (or Adam!) could have hired an actual real, live illustrator to properly make his concept. Everything would have been legal, above board and honestly, better.
But Adam wants people to think HE can do this stuff.
Let’s look at some more…
This is a cool mashup of Poker Face & Columbo, referencing the iconic poster from 'The Sting'. There are a few things that feel off here BUT when you're doing a pastiche like this there's always going to be some weirdness that creeps in. Especially making two different characters in different clothes, matching the pose and style of a different scene. So by all accounts Adam’s done a good job here (don’t look at the table legs though, yeesh)
THEN I found Adam’s first pass...
Woof! This thing has got AI aaalllll over it.
The nonsensical writing on the can, Natasha Lyonne mysteriously blending into Peter Falk’s arm. Columbo’s pants are phasing into the chair. Either AI is doing moajority of the work here or Adam Photoshopped this after pounding 6 ice cold cans of ‘Cooofo Drokee LIGEET’.
You can see here that Adam fixed a bunch of stuff which helped smooth things out (not least, Columbo’s neck) but it still has all the hallmarks of an AI image.
Sticking with the wonderful Natasha Lyonne and Poker Face, here’s a figure Adam made!
Figure looks good! But that artwork…
The garbled text on the hat, the hair phasing through the brim, the glasses melting off her face. These are AI’s mistakes. To be clear NO artist doing a painting or even a paint over would mess up the word ‘ARIZONA’ like that. They might not spell it out exactly, make it suggestive or even block in flat color but they would not do this. BUT AI would!
NOW, people pointed this out (as they quite often do) and Adam clarified things
(Note the original post is deleted, this happens a lot. Either for legal reasons or because Adam is aware it looks bad)
This response from Adam is the entire reason I started digging into his work. I personally hate AI generated imagery but if he’s has managed to find a way to use it and is OPEN about that, then what the hell, go for it.
But that’s not what Adam does. He always, ALWAYS states that he creates the artwork 100% himself and goes out of his way to hide the truth.
Why?
So why would he be doing any of this? Why does he want people to think that he’s the next Drew Struzan, Bob Peak or John Alvin…
Because he sells the things he makes! People want to buy cool shit. They want to buy HIS cool shit!
People are much less likely to want to spend their hard earned money on things if they discover that Midjourney/Stable Diffusion/OpenAI/DALL-E/Krea were holding the paintbrush.
Adam knows this, either consciously or subconsciously he knows that admitting he NEEDS AI to generate the artwork to people will have a negative effect on how his work is perceived and how it will sell.
So, he pretends that it’s 100% him. This in itself takes actual effort. Want to see!?
More Proof
I can’t remember when Adam first got on my radar but I know it was because he made something incredibly clever and brilliant, as so many of his creations truly are. I’m throwing him under the bus for his use of AI (and more importantly, his pretense that it ISN’T AI) but credit where credit is due, he has some fantastic ideas..
Adam has deleted almost all of his posts where he mentions AI but there are a handful on his twitter profile
I followed him for long enough that I can remember him waxing evangelical about AI when he first started using it. If memory serves, the deleted post above was him saying how AI was going to make his work so much better/faster. AI generated work was pretty much all he posted about for a period but Adam is involved enough in the creative communities to sense the backlash to AI and while I’ve not seen him be outspoken about it, he has at least quietly deleted all of those posts where he sung its praises.
But the internet can also remember stuff! Using the Wayback Machine to view Adam’s twitter gives you an insight into the sort of things he was posting then
Sometimes Adam would use AI for quick ‘meme-level’ jokes
That Jack Torrance and Freddy have clearly got the ‘AI’ weirdness to them BUT it’s a good idea! It’s funny! It’s just much, much quicker (and being honest, almost certainly better) than if Adam did them by hand.
I mean, this is 100% FreddAI Kruger. Just look at the book he’s reading AND he doesn't have knives on his right hand!
He still uses it from time to time
These are fun, throwaway ideas so it does feel mean to pull them up, but they do prove that Adam has been using AI. He is familiar with how generative prompt based and style transferred imagery works.
Same things with this series of Fright Night posters
The person who can render those cloud faces (digitally or otherwise) SO well would not fuck up the Exorcist house that badly. That building is in 0.5 point perspective. No wonder Satan moved in.
I know I sound mean but I'm not mocking Adam's lack of skills, its the AI that got it wrong. Once I started looking into it though I was floored by how much of Adam's output over the last year is clearly AI generated but pitched (and sold!!) as original artwork...
If he’s asked about it, he ALWAYS says he painted the artwork himself.
Once again it is SO difficult to be this good at SO MANY different styles of drawing and painting. Digital tools have made it easier but if you have the skills Adam is claiming to have then he makes a LOT of weird, basic errors which seem to contradict this…
He seems to struggle with how foreheads work
And apparently he CANNOT paint anchors
Again, if you’re going to create the image on the left, to that level of finish you HAVE to notice that you’re not painting the anchors properly!?
The same here
The physical figure of the mayor has the tiny anchors hand painted onto their suit but the backing card artwork has some weird nonsensical scribbles. Why would Adam paint them that way? It’s because he didn’t, AI did. And what’s more, he doesn’t have the chops to spot or correct the mistake.
EXCITING UPDATE!
Either Adam saw this (I’m pretty sure he saw this) or he finally learned how to paint Anchors… Check out this new version
I mean sure the anchors on the suit are all copy>pasted in pretty much the exact the same perspective and scale but at least they're anchors now! Great job Adam!
The same thing goes for this poster Adam ‘painted’
The pattern on this carpet is legendary. Zoom in and look at how Adam resolved it here… There is NO way you would compose and paint this scene AND fuck that pattern up this badly. Unless, you didn’t paint it.
Even when he’s posting NON artwork stuff, like this shot of Jamie Lee Curtis holding up the Everything Everywhere All At Once figure he made on Late Night TV (cool!), but look closely at it…
Adam has used AI to upscale it and accidentally create a nightmare in the process. The text on the pack, the likeness even James Corden’s face looks like Andy Richter now… Also he hasn’t noticed any of these things, which is extra weird to me.
Another elephant in the room here is that Adam’s figures often don’t have the best likeness of the characters.
He’s tapping into that 1960’s-70’s Kenner style figure that don’t always need to look anything like the person or character, so the pack does a lot of the heavy lifting. Which again, is why using AI to handle this part helps take his stuff to the next level. However, once again, he doesn’t want to publicly admit this.
And again, if ever he’s confronted about it, often by ‘fellow’ artists, he blocks them
So, what did Adam do before AI assisted image editing was even an option? Well, for a while he used filters/apps, many of which are precursors to to current AI image generators, offering style transfers and emulation, often with jarring artifacts. Before that, he simply used raw photos with some, frankly horrible effects over the top.
Don’t get me wrong, people get better! The more you paint and draw the more you improve BUT these aren't paintings or original artwork they're filtered images. The kind of work students do when they first fire up Photoshop. The gulf in quality between these and the artwork Adam now makes for his pieces is astronomical.
I'm guessing that once Adam got a taste for AI he just couldn't go back. It made his pieces, posters & prints look better than he ever could alone and people loved it. Now he completely relies on it, yet still pretends it's all his own work because the truth would be too damaging to his perceived reputation.
Quick aside, that Normal Rockwell Pee-Wee Herman image was made and sold just after Paul Reubens passed away. I don’t have records to show if Adam donated to proceeds to a charity but if he didn’t then that is the worst kind of opportunistic grift.
Again Adam makes a living off of this stuff. That, by default, makes him a professional. He also goes to extremes to cover his tracks. The more I looked into his practices, the more weird stuff I found…
Down the rabbit hole
Let’s assume that Adam’s painting skills HAVE improved, perfectly in line with the advancement of image filters and then AI generated imagery. Working digitally, it’s hard to prove that you actually painted something. Most artists don’t record timelapses of their work because, well, they’re working! It takes longer, it can make things lag, you start to question your process and in most cases what you’re working on can’t be shared with the public. However, if you work traditionally, with pencils and paints then it’s very easy to prove you made the original, because you physically have it in your hands!
Let’s take a look at this artwork Adam painted for his ‘David’ piece.
Now this could be painted digitally BUT Adam shared a photo of the original, so we can see that he painted it with actual paints!
You can clearly see the brush strokes and the rough painterly edge where the canvas or card is peeking through. No question this has to be a real painting, right?
Of course, none of the visible brush marks in this image match the actual shape of the strokes that would have made them. They run across colors, edges and forms seemingly at random. But hey, maybe its a sloppy varnishing job. Adam scanned it or took a photo and it brought out the varnish in such a way to make it look like someone has overlayed fake strokes in Photoshop. And you can still see the canvas! Right? Even though, the region at the bottom of the painting has multiple different colors and shapes, which wouldn’t leave this sort of area. In fact, the button on the shirt on the pack doesn’t show any of that white canvas, so I’m guessing it was applied after the fact… Maybe to help suggest the idea that Adam painted it and create an illusion of artifice in an attempt to add credibility to his work?
STOP. It’s very possible he did paint this. He just found a photo of Michael Fassbender as David, and painstakingly recreated it with physical paints, matching everything up to the original source photo 1:1
The truth is that this is VERY clearly that photo with a bunch of filters on but Adam wasn’t happy with that, he wants people to think he actually painted it! He’s a PRO artists, not a con artist!! The people he admires don’t cheat this stuff so he’s not going to either. By the only way he knows how, cheating!
Let’s see if Adam’s got any more paintings lying around.
Look! Here are some old paintings & sketches Adam thought were lost! Proof!
I mean it is a little weird that his Toni Colette painting STILL has masking tape around it and that masking tape doesn’t have a single dot of paint on it, but hey maybe that’s just how painting works, for Adam.
It is also a little weird that when he shared the original painting in 2020 it didn't have that tape on it.
And it’s also a different crop.
It seems that while it was ‘misplaced’ somebody painted more of it, accidentally removed Adam’s signature AND put masking tape around it!? A very annoying and definitely real thing that happens to artists!
I’m not even going to get into those two drawings of Stephen King and David Lynch. They are not only screamingly AI but more importantly, if they’re four or five years old as Adam’s post suggests, why do all of his images from that era look like this?
Another big red flag for Adam’s recent work is that in the past, he was heavily reliant on source photos for his ‘paintings’. This isn’t uncommon, Drew Struzan, Norman Rockwell and plenty of other heavy hitters of illustration rely on source photography and as demonstrated above, Adam’s artwork from the time matched up 1:1 to those stills. The thing is, in Adam’s recent work, there aren’t specific photos that he’s using. As well as the style, the angle, perspective and lighting changes drastically. This is NOT easy to do!
This was even brought up in his recent podcast interview. One of the hosts interviewing Adam said:
“Your ability to capture likeness is fucking incredible. I know, as a former artist how difficult that shit is, but you’re not only doing that. Like in your non-sculpt work you’re also capturing these licenses [sic] and working in multiple styles… it’s difficult to capture likenesses and it’s enormously difficult to be able to do that in multiple styles”
To which Adam replied
“With that, especially with the likeness stuff, I’ve started… maybe the last year and a half now… I have so many digital sculpts saved of these characters I’ve made. So I was trying to incorporate that into doing the paintings. So, I’ve been using 3D programs and like Maya to pose them and get the lighting down, like as a reference image… it’s just been an easy way to kind get a good reference for it… and be able to get the lighting down. That’s the most recent thing, maybe about a year and a half I’ve been doing that and I think it’s been going pretty well”
To which one of the hosts follows up with
“That explains a lot actually because one of the things I really like about your work is that it’s clearly not just off a well known promotional still… A lot of these artists can do these incredible things but it’s literally them just recreating a frame…”
As Adam himself admits, this has been ‘a recent thing’ he’s been doing for about the last year and a half. Weirdly coinciding with when he first started using Midjourney BUT WHO IS TO SAY if that’s what he’s doing.
Granted I’ve been a professional toy designer for almost 30 years and have created artwork for commercial toy lines, promotional posters and editorial illustrations, from traditional airbrush work to oil painting and digital and it’s MY belief that Adam’s answer is 100% a lie. None of his sculpts are at a level where he can do what he’s suggesting and EVEN if they were it takes extreme skill to turn even a comprehsnive, well lit 3D render into a convincing illustration in multiple styles.
He has thought about how to hide his true working methods and decided that enough people won’t understand the process he’s suggesting that this will suffice. Not cool!
Just one more thing…
Digging into Adam’s work has been revealing for me. I keep unearthing very strange things he’s done to try and disguise his working methods. Here’s just one (ok, maybe two) more things I noticed.
This is a Weird Al/Daniel Radcliffe figure Adam made back in 2022. I noticed that he posted it again in 2024, but it was different. The photo itself is exactly the same but the figure has been altered in post, either in Photoshop (or with AI’s helping hand).
To me this is most likely Adam being unhappy with how the original turned out, so when he posted it again he couldn’t help but tweak it. Once I clocked this, I noticed that other faces for figures he’s posted have some odd AI looking artifacts, almost always in the hair. I won’t go into these (yet) but this Weird Al example is especially interesting to me as it’s one of the rare times where Adam actually posted his ZBrush sculpt
Look at all the work he’s put into that hair! It’s seems a shame that he wasn’t happy with it. It’s not the sort of thing you can just search a 3D model site for and download.
Typing ‘curly hair’ into a a popular, ‘3d print ready’ catalog site doesn’t give you anything that looks like Adam’s hair sculpt
Wait a minute…
Hang on, the front of the hair is all different in Adam’s version. He must have spent a long time adjust and tweaking it.
I mean, there’s no way he just duplicated it twice, rotated and and scaled those copies…?
Oh. That’s exactly what he did!
But he CAN sculpt things! Take a look at this Trump likeness he made
It’s really good!
Yes, more cynical people than me would point out that there is a 3D model of Trump's head you can buy which looks almost identical in every single way but who is to say if that’s exactly 100% what Adam did here.
Ok look, I am now being unfair to Adam. He can model shit however he wants and commercial artists take shortcuts all the time. He also puts a lot of time and effort into cleaning, preparing and painting his figures - he’s not sitting in a hammock getting AI to print money for him BUT he DOES make money from his work.
The bottom line is, Adam Perocchi is running a business (and to reiterate, it IS a business) and then willfully, painstakingly going to lengths to hide his working methods & create an illusion of skill which benefits him financially. That’s a grift!
I’m sure Adam has found ways to justify all of this to himself but I, personally, think it sucks. It’s not only unfair on the people buying Adam’s ‘original’ art, it’s unfair on actual trained artists whose livelihoods depend on commercial projects - we’re struggling to compete with AI anyway and having ‘artists’ out there hiding behind it, pretending it’s all their own work, willfully, actively lying about it and churning this stuff out at an unachievable human rate, muddies the water even further.
So, if you or anyone you know is ever tempted to bid on a ‘Readful Things’ piece, please think about all of the above before you do.
And Adam, if you’re reading this, this WILL all come back to bite you in the ass. Whether you believe it or not. I guarantee that majority of the people who are giving you their money for the things you make are NOT cool with AI and even less cool with being lied to.
What are your opinions of Readful Things lately? I just got into him and was bummed when I read this