Face filters and selfie apps are so compelling because they simulate limitless interest in what we look like.
Last year ended with images flooding Instagram, Twitter and group chats — pictures of us, but not exactly. They were products of an app called Lensa AI, which can apply machine-learning tools to scan your selfies and generate portraits of you in a variety of artistic styles. This sort of thing seems to be a semiannual trend: Every few months, an app emerges to collect photos of your face and manipulate your image, reflecting and refracting it back to you anew. There have been tools allowing you to apply virtual makeup to selfies, to apply “art filters” imitating famous paintings, to morph into an animal. You could age yourself three decades; you could see yourself imagined as another ethnicity; you could swap your gender; you could become thinner, or bald.
Lensa’s process is fairly simple. You upload 10 to 20 selfies and pay a few dollars for a pack of “magic avatars,” selecting a set of artistic styles from a confusing range of options. (These include “Fairy Princess,” “Fantasy,” “Stylish,” “Light,” “Iridescent,” “Anime,” “Pop,” “Cosmic,” “Focus,” “Kawaii” and others.) Then you wait a few minutes for the app to spit its results at you, in a form that resembles trading cards: pictures that feel as though a whole selection of illustrators and commercial artists have spent their time crafting drawings and paintings of you, for movie posters or album covers or animated films about your life. (The technology involved was, after all, originally trained using huge volumes of images pulled from the internet, including real artists’ work.)
Digital tools can function like fun-house mirrors, feeding a wholly private fascination with ourselves.
I spent $5.99 for a pack of 100 magic avatars, working from selfies I had taken over the last few months. (What was the original point of these selfies, anyway? In one, I’m showing off very red lipstick for friends; in another, I’m holding up a coffee cup and demonstrating that I’m tired; in most, it’s unclear what I’m doing or why I took them at all.) When the avatars appeared, I was startled: There I was and wasn’t. There was something unsettling and halfway-human about them. Some resembled me almost exactly, albeit me in, say, a golden headdress; others transformed my eyes into someone else’s, or gave me a nose that was not my nose. The aesthetics, no matter the chosen style, seemed anime-inflected, with a dose of photorealism. The portraits were, to my mind, ugly and strange, but there was no denying a certain amount of freakish accuracy. I couldn’t stop looking. I had perhaps never experienced the uncanny so directly, or at least in such direct relation to myself. I stared at a “Fairy Princess” version of me, crowned with flowers, eyes averted and lips pursed. Was that really me? Of course not. But also, maybe, a little bit, yes.
One thing about inhabiting a face is that we can never quite see it the way others do. Mirrors give us a reversed image. Photographs freeze us in time at odd angles and, sometimes, in pitiless detail. Staring into a phone camera, preening to check our makeup or undereye circles, gives us a hyperreal mirror, but this, too, is distorted and reversed. Even video doesn’t quite capture us for ourselves, for the simple reason that we cannot watch ourselves objectively. We are looking too closely; maybe we are critical, or maybe we are grateful that we turned out to look a certain way. We cannot simply take stock of how we might look to someone else.
The Rise of OpenAI
The San Francisco company is one of the world’s most ambitious artificial intelligence labs. Here’s a look at some recent developments.
- ChatGPT: The cutting-edge chatbot is raising fears of students cheating on their homework. But its potential as an educational tool outweighs its risks, our technology columnist writes.
- DALL-E 2: The system lets you create digital images simply by describing what you want to see. But for some, image generators are worrisome.
- GPT-3: With mind-boggling fluency, the natural-language system can write, argue and code. The implications for the future could be profound.
This is part of the appeal of portraiture. It is not the objective reality of who we are but a version of how someone else sees us, translated onto the page or the canvas. It always astounds and moves me to think, while in art museums, that before photography, the only simple way to see a static image of yourself was through the brush or the pen or the chisel, necessarily filtered through another person’s creative intelligence. Standing in front of John Singer Sargent’s portrait of Gretchen Osgood Warren and her daughter, we see not just what they might have looked like but the specific things Sargent saw: the high pink of Warren’s cheeks, the dreaminess of her daughter’s expression, the silvery pink of her dress, the candles behind them fading in paint strokes into the background. The way her shawl trails off to the floor, gray and vague, is something only Sargent might have seen and painted.
Cameras have gradually made it unexceptional to see images of ourselves, but there is still something magical about having another person pay this kind of sustained creative attention to you. This is part of why people pay for boardwalk caricatures and commission portraits of themselves in oils. There is real delight in seeing how someone else sees us, experiencing a strange, maybe tenuous connection with their artistic vision. Or, perhaps, with their attention: A portrait, after all, is the fruit of intense aesthetic focus, and what could be more flattering than having that focus fixed on yourself?
Lensa borrows, in an impoverished way, from that appeal. But of course these images are not how another person sees us. They are the naked eye of an inhuman intelligence, combining features mathematically, lips and nose and eyes added up into an approximation of you. There is not, I think, real potential for beauty here — not because the images are produced algorithmically (I believe it is possible for computers to create compelling artwork), but because the avatars seem to me to satisfy neither realism nor artistry. They live between the two and are mostly good for evaluating, once again, versions of how we might look. Their style mimics fantasy characters, comic books and heroes, imagining you into a flattering spotlight — but even in this they fail to elevate you specifically, instead sliding your image into generic visual tropes. You could be anyone, and in fact, you are!
The fundamental appeal of apps like this is, of course, to our own self-involvement. Much has already been said about the way the internet fuels self-obsession by pushing us to perform for others: On Facebook, people announce banal life developments and political opinions; on Instagram, we interrupt our fun to show others how much fun we’re having; on Twitter, we mine our personal lives for laughs. But there is also something to be said about how digital tools can function like fun-house mirrors, feeding a wholly private fascination with ourselves. I can’t help staring at Lensa’s hundred semireal versions of my face. (Actually, 110: There was a bonus of 10 “Holiday Spirit” styles, including one in which I stared down moodily while wearing a Santa hat.) I didn’t share these avatars with others. Even those who did post theirs shared only a fraction of what was produced. These images seemed intended to be marveled at in private. They are like the lists people post of the year’s best moments, or recollections of dreams — things that, even when shared with others, are always more interesting to ourselves.
The fact is that no one has ever before tried to draw or capture 110 versions of my face; no one has ever paid that much attention to what I look like, except probably me. There is much to see. I can scroll and think, here is what I like about me (my eyes, my lips, my hair) and here is what I don’t (my cheeks, my nose, the whole effect seen from certain angles). The app tricks me into feeling seen, but really it is just me, trying once again to see myself.
Source photographs from Sophie Haigney