Camera and lens effects
Phone cameras use wide-angle lenses. Held close to the face, they make whatever is nearest the lens look bigger: the nose and the middle of the face swell while the ears and jaw recede. That distortion changes the proportions the age model reads, and it rarely flatters. Holding the phone at arm's length, or standing further back and zooming in slightly, gives a much more natural face.
Angle matters too. A camera below the face shows the underside of the jaw and any softness there; a camera at or just above eye level is kinder. And because the front camera often shows a mirrored preview but saves an unmirrored photo, the saved picture can look subtly wrong to you, even when it looks normal to everyone else.
Lighting and shadows
Light is the factor most people underestimate. Overhead light, like a ceiling lamp or the midday sun, drops shadows under the brows, eyes, nose and chin. Light from one side carves every crease into a line. The age model reads shadow as depth and depth as age. Soft light from in front, such as a window you are facing on a cloudy day, fills those shadows in and usually brings the estimate down.
Under-eye shadow is a special case. Dark circles and puffiness are strong age cues for people and models alike, and they get much worse under top light. The Dark circles analysis on the same page looks at that area on its own if you want to see how it changes with light.
Expression and pose
- A broad smile adds lines around the eyes and mouth; a relaxed half smile adds fewer.
- Squinting into bright light reads as both older and tired.
- Tilting the chin down creates folds under the jaw; a slight forward lean of the head avoids them.
- A tired face after a long day reads older than the same face in the morning.
- Turning the head shortens one side of the face and can push the estimate either way.
Skin, hair and grooming signals the model picks up
Beyond light and lens, the model responds to the same signals people do: skin texture and tone, the depth of the folds beside the nose and mouth, fullness in the cheeks, the line of the jaw, grey in the hair or beard, the shape of the hairline. Glasses frames, heavy fringes and hats hide some of those signals and add uncertainty. Beauty filters remove skin texture entirely, which is why filtered photos almost always read younger and why unfiltered photos can feel harsh by comparison.
Image quality plays in as well. A sharp, high-resolution photo shows every detail. A blurry or heavily compressed one hides detail, which can push an estimate towards the middle of the model's range, younger for some faces and older for others.
A simple self-test
Pick one thing to test, keep everything else the same, and change only that. Two photos a minute apart, one facing a window and one with the window beside you, will tell you more about lighting than any guide. Put them side by side in Compare, read both estimates and both stability ranges, and look at the quality meter for each. Then try camera distance, then expression.
Treat the results as a guide to the photo, not a judgement of your face. The model is a pattern matcher with real limits; it is less reliable for older adults, children and some skin tones, and the difference between two photos matters more than either number on its own. For the bigger picture, see the face age vs real age page.