How to Take a Good Selfie: The Physics Behind a Photo That Looks Like You

The best selfie is mostly a distance problem, not a talent problem. A phone held at arm's length sits roughly a foot from your face, and at that range perspective enlarges whatever is closest to the lens — usually your nose. Move the camera back, prop it on something, light yourself from a window, and crop in afterwards.

Why a selfie doesn't look like you

Nothing about your face changes between a selfie and a portrait. What changes is the distance from the lens to your nose relative to the distance from the lens to your ears. At arm's length that ratio is extreme, and the camera renders it faithfully. The distortion is geometry, not vanity.

This has been measured. In a 2018 paper in JAMA Facial Plastic Surgery, Ward, Ward, Fried and Paskhover built a mathematical model of the effect and found that a photograph taken about 12 inches from the face makes the nasal base appear roughly 30% wider, and the nasal tip about 7% wider, than the same face photographed from five feet. Nothing was retouched and no lens was blamed. The subject was simply closer to the camera.

The reason is that perspective scales with distance. If the lens sits 12 inches from your nose and 15 inches from your ears, your ears are a full 25% further away than your nose — so the nose projects noticeably larger. Move the camera to five feet and that same three-inch gap becomes a 5% difference, which is a rounding error. The face flattens out into the proportions people see when they talk to you.

Focal length is the usual scapegoat here, and it is the wrong one. Focal length does not bend your face; it only determines how close you have to stand to fill the frame. A front camera is a wide-angle lens — the TrueDepth camera on the iPhone 15 and 16 generation is about a 23mm equivalent — so it fills a head-and-shoulders frame at roughly arm's length, which is exactly the distance that causes the problem. A longer lens forces you to back up, and backing up is what fixes the face.

Two secondary effects come with wide lenses and are worth knowing:

• Corner stretching. Wide-angle lenses stretch anything near the edges of the frame. It is why the people on the outside of a group selfie always look widest, and why you should never let your own face drift into a corner.
• More background, less subject. A wide view drags in the room behind you, which is why so many selfies contain a ceiling, a lamp and half a doorway. A tighter view from further away shows the same head at the same size with a fraction of the clutter.

Newer hardware widens the view rather than narrowing it. The iPhone 17's front camera uses a square sensor that outputs photos at up to 18MP, with a wider field of view and automatic reframing, so you can shoot a landscape selfie without rotating the phone. That is genuinely useful for group shots — but it does not change the physics. Distance is still the variable that decides how your face is shaped.

The fix is distance, not a better phone

The fix costs nothing and requires no new equipment. Every step below buys back distance between the lens and your face, then recovers the framing you wanted by cropping afterwards. A photo shot from five feet and cropped to head-and-shoulders is a fundamentally different picture from one shot at twelve inches.

1. 1Turn the phone around. The rear main camera is the better camera on essentially every phone: bigger sensor, more resolution, better optics. It is not dramatically longer than the front lens — an iPhone main camera is around a 26mm equivalent against the front camera's 23mm — but most phones offer a 2x view from that sensor, and more importantly, using the rear camera means you cannot hold it at arm's length while watching yourself. That constraint is the point.
2. 2Get to three to five feet. Five feet is where the face stops distorting. Three feet is a workable compromise. Twelve inches is the failure case.
3. 3Give the phone something to stand on. A stack of books, a shelf, a windowsill, an inexpensive tripod. Anything that holds it steady at the right height. Propping the phone is what makes distance possible at all.
4. 4Trigger it hands-free. Use the 3- or 10-second timer, a smartwatch remote, or a burst. On iPhone you can swipe the shutter button to the left to fire a burst, or turn on Settings > Camera > Use Volume Up for Burst and then hold the volume up button. Either way you shoot dozens of frames without touching the phone between them.
5. 5Frame wider than you need. Leave room around your head. Modern main cameras produce files far larger than any profile picture requires, so cropping in costs you nothing visible.
6. 6Crop last. Do the head-and-shoulders framing in the photo editor, not with your arm.

If you genuinely have to shoot a front-camera selfie — no surface, no helper, no time — extend your arm fully, hold the phone slightly above eye level, and keep your face in the middle of the frame rather than to one side. That combination gets you most of the available improvement in about two seconds.

Light: the variable that outranks the camera

Light does more to a photograph of a face than any camera setting. The only variable that really matters is how big the source appears from where you stand: a large, close source wraps around your features and softens the edge of every shadow, while a small or distant one carves hard lines into your face.

That is the whole principle. The sun is enormous but about 93 million miles away, so at noon it behaves like a pinpoint and cuts hard-edged shadows. A cloud layer turns the entire sky into the source and the shadows go soft. Indoors, a window is the largest source you own — and the closer you stand to it, the larger it appears from your face, and the softer the light becomes.

What to do with a window

Face the window, or turn about 45 degrees away from it. Facing it straight on gives you flat, even, forgiving light with catchlights in both eyes. Forty-five degrees gives you shape: one side of the face brighter, a gentle gradient across the nose, a sense of dimension. Stand three to six feet back — closer for softer light, further for a bit more contrast. Turn off every lamp in the room while you do it, because mixing daylight with tungsten bulbs makes correct skin color nearly impossible to recover afterwards.

Why ceiling lights ruin faces

A recessed downlight is a small source directly above you. Small means hard-edged shadows; above means those shadows land in your eye sockets, under your nose and under your chin. The result reads as tired and slightly hollow, and no amount of editing puts light back into an eye socket that never received any. If overhead light is all you have, tilt your chin up a degree or two so the sockets catch some of it, or better, move to a window.

Why the flash flattens you

The flash is a tiny source sitting on the same axis as the lens. Because the light travels along your line of sight, every shadow falls directly behind you where the camera can't see it. Shadows are what give a face depth, so removing them removes the modeling: the face goes flat, the skin goes shiny, and the background falls away into black. On the front camera, "screen flash" replaces the LED with a glowing white screen — a slightly larger source, still on-axis, still flat.

A quick diagnostic pass, once you see a frame you don't like:

• Dark eye sockets → the light is too high. Lower it, move to a window, or raise your chin slightly.
• A hard-edged nose shadow → the source is too small or too far. Get closer to the window, or diffuse it with a sheer curtain.
• Shiny forehead and nose → the light is very direct and your skin is reflecting it. Blot, or turn slightly so the source hits at an angle.
• Your face darker than the background → you're backlit. Turn around so the window is in front of you.
• Skin looks orange on one side, blue on the other → two light sources of different color temperatures. Switch one off.

If you want the full explanation of key, fill, direction and shadow ratios in plain English, our guide to headshot lighting covers the same physics for a deliberate portrait setup, where this article is about making the best of whatever light happens to be in the room.

Camera height, chin and shoulders

Once distance and light are handled, small geometric changes do the rest. Camera height decides whether the viewer looks slightly down at you or up your nostrils; chin position decides what happens to your jawline; and shoulder angle decides whether you read as a passport photo or as a person.

Height. Put the lens at eye level or one to three inches above it. Slightly above works because it opens the eyes toward the camera and puts a clean line under the jaw. Go much higher and the head grows and the body shrinks into a cartoon. Go below eye level and you are shooting up the nose, compressing the neck, and looming the jaw — the single most common flaw in phone self-portraits, caused almost entirely by holding the phone at chest height.

Chin. Push your forehead slightly toward the camera and drop your chin a fraction. This separates the jawline from the neck and removes the soft shadow under the chin that reads as a double one. It feels ridiculous and looks correct — which is the general rule for almost anything that works in front of a camera.

Shoulders. Square shoulders facing the lens read institutional. Turn ten to thirty degrees off-square, then bring your eyes back to the lens. That small rotation narrows the shoulder line, adds a diagonal to an otherwise symmetrical frame, and looks like you were standing there rather than posing.

None of this is posing for a formal portrait, which is a bigger subject with its own rules. If the goal is a proper headshot rather than a good casual photo of yourself, our guide to posing when you're not photogenic is the one to read — it covers what to do with your hands, weight and eyeline across a full session. And if you wear glasses, the tilt that kills reflections is a specific move worth learning from our guide to glasses in headshots, because a window is exactly the kind of big, bright source that lands as a white rectangle in your lenses.

What's behind you is half the photo

The background is roughly half the pixels in a head-and-shoulders frame, and viewers read it whether or not they notice it. Three things matter: how far it sits behind you, whether it competes with your face for attention, and whether it separates your hair and shoulders from the wall.

Distance is the lever again. Standing with your back against a wall gives you a flat, evenly lit surface with your own shadow printed on it. Step three or four feet forward and the wall goes softer and darker, your shadow drops out of the frame, and you get separation for free — no portrait mode required.

• Tone contrast beats color contrast. Dark hair on a dark wall merges into a silhouette. Choose a background a few steps lighter or darker than your hair and jacket.
• Watch the verticals. Door frames, poles and picture edges that appear to grow out of your head are invisible while shooting and impossible to ignore afterwards.
• Movement in the background is noise. A bookshelf is fine; a bookshelf where a viewer starts reading titles is not.
• Check the corners before you shoot, not after. Two seconds of scanning saves the whole frame.

Two of our articles cover this in more depth and they do different jobs: the headshot backgrounds guide is about choosing a background for a given purpose — corporate, creative, academic — while the piece on LinkedIn profile picture backgrounds deals specifically with what survives a small circular crop, which eliminates most backgrounds that look fine at full size.

Shoot thirty frames, not one

Stop trying to produce one good photograph. Expression is a moving target, and the difference between a warm face and a stiff one is frequently a single frame — nobody can hold a natural expression on command while also pressing a button. Shoot thirty frames and select afterwards.

Working in bursts also solves the problem the timer creates. When you know a shutter is about to fire in three seconds, you brace for it, and bracing shows up in the eyes and the corners of the mouth. When the camera is already firing continuously, there is no moment to brace for, and the frames drift toward something ordinary and relaxed.

A method that works:

1. 1Fire ten frames you fully intend to delete. They are there to get the self-consciousness out of your system.
2. 2Reset between bursts: look away from the lens, exhale, then look back. The frame immediately after the return has genuinely fresh eyes, which is very hard to fake.
3. 3Change one thing per burst — chin, shoulder angle, distance, expression — so you know what caused a difference.
4. 4Cull on your phone at thumbnail size, not full screen. That is closer to how the photo will actually be seen.
5. 5Judge the eyes first, everything else second. A photo with the right eyes and an imperfect mouth beats the reverse.
6. 6Pick three finalists, then look again an hour later. Your first choice is often the one where you look most like your idea of yourself rather than most like yourself.

Whether you smile — and how — is worth its own decision rather than a default. The mechanics of getting a real one, as opposed to a held one, are covered in our piece on smiling in a headshot.

Portrait mode and beauty filters, honestly

Both features change your face, and only one of them is usually worth leaving on. Portrait mode builds a depth map and blurs what it decides is background; beauty filters detect facial landmarks and rewrite them. The first fails at edges, the second fails at texture, and texture is what makes skin look real.

Portrait mode is a reasonable tool with a known failure mode. A phone is too thin to contain a lens that blurs a background optically, so it estimates depth and applies blur in software. Camera test labs evaluate exactly this: the quality of the simulated blur, and the artifacts produced when the phone tries to isolate a subject from its background. That estimate is good at torsos and bad at fine, semi-transparent edges — flyaway hair, curls and coils, beards, glasses frames, and hands near the face. When it fails you get a chunk of blurred hair, a halo, or a hard line slicing through soft blur. It also degrades in low light, because the depth estimate needs signal to work with. On phones with multiple rear cameras, portrait mode typically shoots at a longer equivalent focal length than the standard wide view, which — per everything above — is flattering on its own merits. Use it in good light, check the hairline at full zoom before you keep the frame, and remember that real distance from a real background achieves the same separation with no artifacts at all.

Beauty filters are a harder no for anything professional. Smoothing works by removing high-frequency detail — pores, fine lines, the micro-shadow that tells a viewer skin is a surface with depth. Take that away and the face reads as plastic even to people who cannot articulate why. Viewers detect the absence of texture long before they detect a specific edit, which is why over-smoothed photos tend to feel dishonest rather than flattering. Google reached a similar conclusion from a different direction: starting with the Pixel 4a and Pixel 5 generation in 2020 it shipped face retouching off by default and dropped "beauty" from the labelling, citing its own studies and consultations with child and mental-health experts about filters people don't know are running.

There is a research literature building here too, and it is worth quoting narrowly rather than broadly. A 2024 experiment in Computers in Human Behavior found that exposure to beautifying augmented-reality filters applied to a viewer's own face reduced body and face satisfaction, with no equivalent effect from "uglifying" filters, and with the effect weaker among women with higher self-esteem. A companion argument in Topoi the same year frames filtered self-images as distorted mirrors that can gradually normalize the edited version of your face. Neither is an argument about photographic quality. Both are a reason to be deliberate rather than default about leaving smoothing on.

We build an AI headshot product, so we will say the obvious thing plainly: AI-generated portraits have their own version of this failure. Over-smoothed skin, wax-like texture and mangled hairlines are exactly what makes a generated headshot readable as generated, and we catalogued the tells in our breakdown of the ten signs an AI headshot looks AI-generated. A photographer with a real lens in a real room does not have that failure mode, which is a genuine point in their favor. The standard is the same regardless of how the photo was made: a face should still look like a surface.

Which guide for which job

Selfie craft, headshot posing and input photos for an AI generator are three different jobs with three different success criteria, and confusing them is why people end up with a technically good photo that is wrong for the use they had in mind. Here is which guide covers which.

• This article — general self-portrait craft with a phone, for any use: a dating profile, a Slack avatar, a photo for your parents, a decent stand-in profile picture.
• Feeding an AI headshot generator — a completely different brief, because the model needs varied angles, varied lighting and honest, unfiltered input rather than one well-composed frame. Our guide to input selfies for AI headshots owns those requirements, including how many photos to supply and which ones actively degrade the result.
• A formal headshot session — where posing, wardrobe and expression are planned rather than discovered. Start with the posing guide and the wardrobe guide.
• A full DIY setup at home — if you want to go past a good selfie into an actual studio substitute with a proper light, a backdrop and a repeatable workflow, our home headshot method is the long version.

The ninety-second version

If you only remember one sequence, use this one. It takes about ninety seconds, works with any phone made in the last five years, and removes the three failure modes that account for most bad self-portraits: too close, lit from above, and only one frame taken.

1. 1Find the biggest window in the building and turn off every lamp in the room.
2. 2Stand three to six feet from the window, facing it or turned about 45 degrees away from it.
3. 3Step three or four feet away from whatever is behind you.
4. 4Prop the phone on something at eye level or an inch or two above, rear camera facing you, three to five feet away.
5. 5Set a timer or start a burst, then fire thirty frames — resetting your eyes between them.
6. 6Crop to head-and-shoulders afterwards. Adjust exposure if you must. Leave the smoothing slider alone.

Everything above is craft you own permanently. It costs nothing, works on any phone, and it is the same physics a photographer is applying with better equipment in a better room.