Everything Wrong with AI Art
Everything Wrong with AI art in the Concept Art Field and Video-Games Job Market
Commercial work has always paid the bills, but concept artists have, in large measure, have it dictate what they do.
Character skins, loot boxes, prize crates, your skills have been employed to illustrate things that were unessential at best.
The market has always rewarded this behavior, job listings and concept art awards of many kinds, reinforced it: that the best use of our skills was to design content to be consumed fast, fatten the board of director’s wallets, capitalize on what consumers buy, and already project what comes next.
We, as concept artists, have always bent over our artistic process just enough to meet demand, faster work was more rewarded than better work. I remember I'd be at school and being told that to design things well, I should travel, research books, talk to professionals of the field I was looking into (if you’re designing a spaceship, talk to a scientist); where do we get our reference today? Google Images, Pinterest albums, Star Wars movies.
(To be frank, this is only 90% true. This article almost was going to be about "Everything Wrong with the Entertainment Industry" but we’ll leave that for another time.)
Moving on now, I suggest a shift of priorities, and we move from being artists into being designers and storytellers.
I've never been a fan of pop corn entertainment, such as the movies of the past decade, and the same goes for games.
The predictability of the plot and how it's always been crafted not by creatives, but by layers of paid executives, accountants, who'd vet the scripts in case they were not dumb enough to cater and amuse the lowest common denominator. Such end product really interested me. Not because I’m smarter, but because it doesn’t make me relate to it, it doesn’t make me feel anything.
What have we, as concept artists, done that wouldn't be generic enough on the job to be replaced by an AI automated process of any sort?
“Art is not what you see, but what you make others see.”
Here's what I suggest, a new shift of priorities, and what can't be automated:
Design is not an innocent practice, after almost a decade of working in the games industry, we always heard the meme of "Concept Art is not about creating pretty pictures" - it's now more true than ever.
At the freelance concept art brand I head, Menog Creative Crew, I've made it a regular practice to exercise graphic design in my work, this forces me to think in ways AI is not and frankly, will never reach human heights. The act of design is to bridge a gap between the audience, clients goals and market needs, to channel that into the product and establish ways of communicating with said audience. It aims for something that hasn’t been done before, and therefor, it’s not mapped in a database which it can access to.
AI can certainly build on and amplify on the meme: it can create visually stunning artwork, but it lacks the social awareness for an audience niche, intricacies of clients needs and the changing trends and it will struggle to adapt to what's out of its database scope. It does great at playing fetch and which serves that it is just as smart as a dog, but it has absolutely no clue of what it’s fetching, it’s hardly intelligent, but does a great job at deceiving you to think otherwise, just like a dog doesn’t recognize what a newspaper actually is, no matter how much it is trained to get it for you and bring it to your lap.The solution we propose at the brand: An OKR form is a good start. While I've made my own form to fill with clients, the point is to understand that during discovery, it's crucial to find how to construct designs that will communicate with the audience in a way that establishes a dialogue that they want to have and not let go of.
Andrew Groove, teaching the OKR method at Intel, during the 70’s.
“Creativity is not about getting ideas, but it’s about getting ideas to work.”
Think about the time you played your favorite game, and what made you so into it, and take the time to think about how similar were you to the friends and family who also enjoyed it, how much alike are you? You likely share the same persona profiles in some way. You’re the same target audience and demographics, you’re the client you’re designing for, in some way.
With me, it was the Fallout series, specially Fallout 3 and New Vegas. As a young boy who wanted to escape from life's problems, I needed a safe haven that was relatable to my grim, but still have fun and carry a wildly humorous beat, a glimpse of hope. It was hopeless but absurd, in a positive sense, all was meaningless but the characters and the world kept finding new meanings to live from. Emotions resonate with people and one thing we can all agree on, is that an AI will never be able to do, to feel, and to ultimately discover ways to relate to people by genuine feeling, that's what we're selling as concept designers. We design, pitch, and sell the experience, the meaning, gamers are in it for the journey.
You can try to express your feelings about anything to anyone, but the way we go about it, is what sells our own unique voices and authenticity, which has the power to turn us in or out. There's always a need in the market to be fulfilled, a void to be replaced, a listening device to be built to relate to someone. A story to be concepted.
At Menog Creative, we're the bridge builders, our raw materials are authenticity.
Okay… but, how?
First, forget about good, good is a known quantity, as long as you focus on good, you'll never have real growth.
Throw away the ornaments and the desire to impress by making a good artistic render. Forget about fattening up your wallet for a moment and focus on delivering something people want, then, you’ll get revenue, because people will want to pay for it, out of sheer need and want, not out of impulse or addiction or persuasion.
Focus on what others want and need, this isn't about you. You're providing a service.
If you're doing a space game, for example, do it because you have a genuine and entertaining story to tell and people need to play it, because it's grounded on an emotional level on something a set of people will relate and engage. Maybe you, like Sparth, moved overseas young and spent your time looking at airplanes and early spaceships, in awe. Well, now you have a feeling to pull from, and a story will set in place.
Concept art by Sparth for the HALO franchise.
(Don't do it because you want to push your game engine, or game design, or whatever other nerdy things your audience doesn't care about, your buddies at GDC will be impressed, your player-base won't.)
Let’s do some big thinking here together and come up with a project on the spot: mock up a skate game, you and I. For this, we have to know what skate is, first and foremost. An AI cannot know what skate is, AI has never skated, AI can only know what skate looks like, it can get reports of what skate feels like, to the broader audience, but it cannot know what skate means to you and me, both.
If you go to Google Images or Pinterest, and look for a “skater” this is what it will give you:
Carefully picked and clean fashion. This isn’t skate. Skate is raw, it’s like active meditation, it’s freedom of movement as well as freedom of emotion. This is just something they (the store where I got the picture from) want to use to sell clothes, and which appeared in the first page of Google images to me.
I’m unsure of what this is, but I can guarantee that I’ve never seen someone skate with pristine shoes like that.
Emo sadboi archetype with skinny jeans, yes, definitely not the usual skater you’ll see with your own eyeballs, or speak to.
There’s the worst sin you can commit in approaching this theme and topic from the lens of an artist…
Settle for poser.
I’ve seen it happening, even with AAA IP’s. Of course, most don’t notice, but this on the left, isn’t someone you’d see skating anywhere in the real world, maybe in poser-land you would, but this isn’t a real character design reference. This isn’t reference material, and it’s the culprit of letting Pinterest curate reference for you.
In turn, this is an AI image I found online as well:
Classic, corny, video-game like environment for a skate game (full of graffiti); with fashion design that looks like something out of the 8 mile or another rap movie. I mean, it does fine at selling to an audience that doesn’t skate, doesn’t understand, doesn’t relate, and just seek to be amused. This project wouldn’t be loved nor hated by anyone, but it’d likely turn some profit. There are colors, there’s some mood in the Google image reference, there’s some understanding of design coherence by the AI as well. At the end of the day, though, this project would not be loved nor hated by anyone. It’s the collective conscious understanding of skate. If an AI replaces you with this, then I’m sorry to put it so raw but, you deserve it.
The best way to research is to get involved, relate to the subject matter, like you were establishing a relationship with them.
What does good reference look like? Scroll down.
I’ve actually been involved in a skate community, I’ve set up my graphic design and marketing skills to use of the local skate and surf culture, I wanted to have a project that was just about giving back to someone or something, and at the same time, that I could live at. It’s been an enormous gift to be able to have such a project in my life where I could practice altruism, but taught me many things as well. The subject matter, being skate in this blog, was not an accident, as you may realize by now. I’ve become acquainted with the subject, I am, a living research subject for such a project, if I was to ever take part on a skate IP. I am confident I can manifest and realize it, as I’ve lived it, experienced it, I know not only what a kickflip or an ollie looks like, I’ve felt it, I know the ins and outs of it. I know the emotions that accompany it before and after, and how the atmosphere of your skate crew changes or doesn’t in accordance to whatever action comes or goes in this dynamic environment.
I guess, if anything, I’m outlining here the importance of research. At Menog Creative, we believe that to take on any project, one should relate to it, get intimate with it, we don’t only manipulate graphics to look vintage or retro, to pay a better homage to skate culture, we get a Super8 old camera and film our reels with it, we photograph our own magazines to make our flyers and announce skate sessions, this is an authentic approach to the creative process, one we believe sets a competitive edge, and we’re eager, always, to employ to any client project.
(Marina Correia on the left)
Without further ado, here’s what I would believe a skater looks like, it can, for sure, vary, everyone has their own style, but it often overlaps in what it doesn’t look like:
Skinny jeans, that’s just impractical.
Emo looks, everyone has a good vibe when skateboarding. I have yet to meet a depressed skate person (unless they are a pro athlete or something).
They worry more about what’s practical to skate, than what someone thinks of them.
If they do have some aesthetic to them, it’s only because strong brands like Vans, Adidas Skateboarding, Supreme, Girl, Deathwish, and others, understand their audience. Audience buys said stuff from these companies, which happen to do amazing designs.
Gustavo Ribeiro, Portuguese pro-skater. Closer to a modern day Oi! Subculture than any other references.
Closer to Kurt Cobain’s grunge than anything we’d seen so far from the collective conscious of Google images, Pinterest and AI.
The takeaway here: it’s important you can read between the lines. You may argue that you cannot tell the difference apart, and that’s why you need a culturally rich, socially aware designer to be able to tell reference apart for your game or film project.
On the reference above I shared after, as examples of good reference, there's something free, wild, there’s this edgy masculine vibe but a feminine gracefulness to it. It’s a ying-yang. Looking back, I would never be able to get an AI to research things for me, let alone if I didn’t know, beforehand, where the common misrepresentations of the skate universe would fall. To find what is it that my emotions say about these archetypes and people in specific, what is it I read from a far, that speaks skate itself. One may argue that an AI can be trained to do these things, but if you do have to train an AI to teach it things to such a minute detail of what skate means to you, then it’s coming closer than being your student than your servant, and may just perform constructively for being something like an assistant, or junior artist who you direct. Point being, is that you cannot teach AI to be you and to replace you, unless what you do is extremely generic and replaceable.
AI would never be able to express my own view, and that’s what connects people and skate crews.
I know AI cannot do it, because google images references can't do it, Pinterest does a so-so job at it and even Tumblr, a more authenticity-centric nest of references, may struggle. This is because these are other people’s interpretations of skate, and they've been reproduced and rephotographed, scanned, copied, translated in many mediums and forms, in ways that have bastardized someone's view of the subculture, its meaning, and emotions.
It’s no surprise if I tell you I approach this, this design, as a love letter about skate. Because I skate myself and have skated for the past half a decade, with no intention of stopping on the horizon, I have, therefore, made my research and became the research.
This is absolutely imperative, and requires a deep pause before a further dive.
Let’s shift our focus to the research process:
The research process is an execution process, it's not a checklist. It's not your mood board, it's not a task, it is the project.
All clients I have worked with were kind enough to trust me with enough time and resources to manifest my work the way I see it, they allow me to direct it. They come to me for my ideas (most of them, anyway)
An AI can not maintain relationships which is a big part of what a designer does, (Like the process of Discovery), the audience insights are not always obvious and require a sort of… therapist-like approach to philosophical and intellectual surgery on understanding what is it, really, that we’re seeing in front of us, what really is this subject about? Not what it looks like, but what it truly is? This is why, concept art isn’t made to just look pretty. (Or isn’t supposed to.)
An AI does not do anything that wasn’t already obvious, nothing that isn’t generic, but no one seemed to bother much: that all cars look the same, theater is dead and films became movies, and movies exist to sell pop corn rather than a story. All AI has done, was to speed up the surfacing of this problem, and reminded us more of it, by exposing it. That the content we consume as an audience needs to be dumbed down enough for entertaining us and numbing us, with vetted scripts by layers of paid executives, lawyers, accountants, analysts, who will decide if the creative work is basic enough to be understood by the masses, and therefor, sell to most the audience possible on Earth. Again, I stress, content to be neither loved nor hated by anyone, but simply digested.
A designer sets trends and foresees horizons, AI simple runs in circles within those horizons.
“If I were to ask you to consider what you would do if you were God, you might find that being omniscient, all-powerful, and eternal would eventually become extremely boring. Eventually you would want a surprise.
What are we trying to do with our technology? We are trying to control the world. We are trying to become all-powerful and omniscient. Imagine the ultimate fulfillment of that desire.
When we are in control of everything, and we have great panels of push buttons whereby the slightest touch fulfills every wish, what will we want then? We will eventually want to have a special, red button marked “surprise” built into the panel. Touch that button and what happens?”
Here’s what I propose, from here on out:
When you have concept environments for an old west IP, go to Grand Canyon, take a field trip.
When you have to concept characters for a horror IP, talk to shady characters, like weird looking fortune-tellers at midnight.
When you have to do a skate project, you don’t have to know how to ollie, but know enough to understand what it is, before trying to (mis)represent it.
Before manipulating graphics to look vintage, try to rent a Super8 or a vintage Kodak camera and shoot your own textures and reference.
The caveat, positives of AI:
Even with all of this, I advocate that we must embrace AI, AI can be a useful tool, AI allows an artist to fail fast and fail many times in a day, without being involved in emotional investment and inner turmoil. AI can help speed up your process.
A few photos of a recent field trip we've taken, at Menog Creative Crew, to capture and sketch new environments and gameplay design. We curate our own reference, we don’t let an AI or Google Images, or others to curate it for us.
Why? Let’s just get this out the way: creativity is not device dependent, and the problem with AI is that everyone has it.
There’s a lot of talk about work-life balance.
I’m more into work-life integration. Field trips for work are like a moving and remote laboratory, where we can engage with the subject matter.
Closing thoughts:
AI also focuses heavily on outcome which predicts results, and you'll know where you will go, when focus on process you may not know where it'll take you, but you know you want to be there.
My ideal clients, now more than ever, are lunatics, people who go out and swim against the stream. People who want to create a future and not just fatten their wallet.
Looking back, what did the golden age of gaming, get us in the entertainment industry, to us Seniors? Game Boy Color in atomic purple skin, Nintendo games, Atari, Capcom street fighter machines in neon covered magical atmospheres. Entertainment design that remains iconic. What do we get today from our AAA industry?
Junk mail, pitches, MMO subscriptions, skins, loot boxes.
Design is about relationships. It is intimate. This is timeless, trying to predict industry trends, and if AI will stay or go is as pointless as an endeavor as those that entertained the idea of flying cars in the 2000s, and Macromedia Flash (who?) taking over the web.
MacBooks didn’t automate or invent new typography, it did not put the font’s market away. It simply reframed things. Whatever you do, unless it’s extremely generic, you aren’t 100% replaceable.
About the author:
If you're seeking a seasoned concept artist with a proven track record, look no further than Miguel Nogueira. Contact him directly to embark on a collaborative venture that will transform your creative vision into captivating visual masterpieces.
Contact here.