A free attractiveness test that scores facial symmetry and proportion from a photo, right in your browser, with nothing uploaded.
What the attractiveness test measures
The attractiveness test at faceshapetest.app is a free, browser-based tool that scores facial geometry from a single photo. It reads 478 points across your face with MediaPipe Face Landmarker, running entirely on your device. The photo is never uploaded to a server.
The result is a face score from 0 to 100, built from four measurements: facial symmetry, vertical thirds, horizontal fifths and the golden ratio. Each one compares your proportions against a classical rule from art and anthropometry, then the four are blended into a single number.
Say this plainly up front: this score measures geometric regularity in one photo. It does not measure beauty, worth, or how someone else sees you.
Research on facial attractiveness finds that symmetry and averageness predict attractiveness ratings only weakly, and only on average across large groups of faces (Rhodes, 2006). Expression, grooming, skin, hair, confidence and context shape real impressions far more than any ratio does.
How the score is calculated
The face score blends four geometric measurements into one number, each weighted by how strongly it tends to relate to perceived regularity. Symmetry counts for the most, the golden ratio for the least, and the two proportion checks sit in between.
| Component | Weight | What it compares |
|---|---|---|
| Symmetry | 35% | 15 mirrored left/right landmark pairs across the face's vertical axis |
| Vertical thirds | 25% | Middle third (brow to nose base) against lower third (nose base to chin) |
| Horizontal fifths | 25% | Each eye width and the gap between the eyes against one fifth of cheekbone width |
| Golden ratio | 15% | Three proportions compared with phi, 1.618 |
Symmetry gets the largest share because it is the measurement most consistently, if still weakly, linked to attractiveness ratings in facial research (Perrett et al., 1999). The golden ratio gets the smallest share for reasons covered further down this page.
All four numbers are calculated the same way for every photo, from the same 478 landmarks, then combined into the single face score you see on your result.
What is a good score?
There is no pass or fail line on a facial symmetry test like this one, but scores do cluster in predictable bands. Most people, in a decent frontal photo, land somewhere in the middle to upper range, not near either end.
| Score | Label | What it usually means |
|---|---|---|
| 85–100 | Very regular | Proportions closely track the classical thirds, fifths and symmetry rules |
| 70–84 | Regular | Solid, typical proportions with the small variation almost every face has |
| 55–69 | Typical | Noticeable but ordinary asymmetry or proportion variance |
| Below 55 | Usually a photo issue | Turned head, smiling, or lens distortion, more often than unusual anatomy |
Most good frontal photos of most people land somewhere between 70 and 90. Human faces are, on the whole, quite regular. None of these bands describes anyone as unattractive. A lower score is a cue to check the photo, covered in the checklist below, before it is read as anything about your face.
Symmetry
Symmetry carries the most weight in the face score, at 35 percent. It compares 15 pairs of left and right landmarks: eye corners, brows, cheekbones, jaw points, nostrils and mouth corners.
Each pair is mirrored across a vertical axis running from mid-forehead to chin. The test measures how far each pair sits from a perfect mirror image, averages that mismatch across all 15 pairs, and normalizes it by your face width so the score does not depend on how close you stood to the camera.
Nobody's face is perfectly symmetric. Everyone has some mismatch between their two sides, and that is normal anatomy, not a flaw.
Research on facial symmetry and attractiveness finds a real but modest link between the two at a population level (Perrett et al., 1999). It is one factor among many, not a decisive one for any individual face.
Facial thirds and fifths
Facial thirds and fifths are classical proportion rules from portraiture and anthropometry, going back to canons documented in detail by researchers like Farkas. They describe an idealized average, not a rule any real face is expected to match exactly.
The classical thirds rule says the forehead, mid-face and lower face should be roughly equal in height. Because the hairline is not a landmark our model can locate reliably, our facial symmetry test compares only two of the three.
It measures the middle third, from brow to the base of the nose, against the lower third, from the base of the nose to the chin. The closer that ratio sits to 1:1, the higher this component scores.
The classical fifths rule says the face is roughly five eye-widths wide. We compare the width of each eye and the gap between them against one fifth of your cheekbone width.
Faces close to that even five-part split score higher here. Faces with wider or narrower eye spacing relative to cheekbone width score somewhat lower on this one component.
Both of these are averages drawn from population studies, not fixed targets. Real, typical faces vary around them in both directions.
The golden ratio, and why we weight it least
The golden ratio, roughly 1.618 and written as phi, is a proportion found throughout nature and used for centuries in art and architecture. Applied to faces, it shows up as claims that an "ideal" face has a specific length-to-width ratio, or specific ratios between features.
Our test checks three such ratios: face length to face width, mouth width to nose width, and the distance from the base of the nose to the chin against the distance from the lower lip to the chin. Each is compared with phi, and closeness to that number raises this component of the score.
The golden ratio face idea, popularized by the Marquardt mask in the early 2000s, is widely repeated online and in some cosmetic marketing. It has thin scientific support.
No large body of attractiveness research treats phi as a reliable predictor of how attractive a face is rated. That is why this component carries only 15 percent of the total, the smallest of the four.
We keep it in the blend because it is a well-known part of how people talk about facial proportion, including in this test's own name. We weight it lightly because the evidence behind it is weaker than for symmetry or the classical thirds and fifths.
How to get an accurate score
A handful of photo choices move the measured score by several points, independent of your actual proportions. Getting these right matters more than anything about the face itself.
- Hold the camera at eye level, not above or below.
- Face the camera straight on, without turning or angling your head.
- Use a neutral, relaxed expression rather than a smile.
- Light your face evenly from the front, avoiding heavy shadow on one side.
- Avoid a close-up phone selfie. A front camera held at arm's length uses a wide-angle lens that distorts proportions, especially the nose and jaw.
- Pull hair back off your forehead, brows and jawline so the landmark model can find them.
Following this checklist gives the model a cleaner, more representative set of landmarks, which produces a more accurate face score. If a result surprises you, retake the photo with this list before assuming something is wrong with the reading itself.
What the score does not tell you
The face score comes from one photo and four geometric measurements. It says nothing about your expression, your grooming, your skin, your hair, how you carry yourself, or how you come across to another person over time. Those factors, well documented in attractiveness research, tend to matter more than any single ratio (Rhodes, 2006).
It also does not tell you where you rank against other people. This tool has no tiers and no ranking scale. That is a deliberate choice.
Some online communities, often under the label looksmaxxing, use rating scales like PSL to sort people into tiers. That is a different rating culture with different goals, and it is not what this attractiveness test does.
The score is also just one reading. A different photo, a different day, a different expression can move the number without your face having changed at all. Treat it as a snapshot of one image, not a verdict.
If you are curious about appearance changes for reasons beyond curiosity about a number, consult a qualified professional rather than acting on a single photo score.
The same scan that produces your face score also reports your face shape, canthal tilt at /canthal-tilt, eye shape and skin undertone, all from the same set of landmarks and the same photo.
Frequently asked questions
What is a good attractiveness test score?
Most good frontal photos land between 70 and 90, because human faces are naturally quite regular. There is no cutoff for "attractive," since the score measures proportion, not beauty. A lower number more often points to a photo issue, like a turned head or a smile, than to anything unusual about your face.
Does a low score mean I am not attractive?
No. The score measures geometric regularity in one photo, not attractiveness, worth or how people actually see you. Research finds symmetry and proportion predict attractiveness ratings only weakly and only on average. Expression, grooming, skin, hair, confidence and context matter far more in how someone comes across in real life.
Is the golden ratio face real?
The idea that an ideal face follows the golden ratio, popularized by the Marquardt mask, is widely repeated but has thin scientific support. It is one classical proportion among several, which is why we weight it least of the four measurements here, at only 15 percent of the total score.
Why did my score change between two photos?
Camera angle, lens distortion, expression and lighting all shift the landmarks the test reads, often by more than your face actually changes day to day. A phone held close distorts proportions through a wide-angle lens. A slight head turn or a smile can move the score several points either way.
Is this the same as a PSL or looksmaxxing rating?
No. PSL scales and looksmaxxing tier systems rank people against each other using informal, often harsh criteria. This tool reports a geometric face score with no ranking language and no tiers, built only from measurable proportions in your photo, not a judgment of where you fall against anyone else.
Can I improve my score?
A straighter, better-lit photo usually moves the number more than anything else, since lens distortion and head angle affect it heavily. Beyond the photo itself, grooming, hair and expression can shift how proportions read. For appearance changes beyond that, consult a qualified professional rather than chasing a higher number.
References
- Rhodes, G. (2006). The Evolutionary Psychology of Facial Beauty. Annual Review of Psychology, 57, 199-226.
- Perrett, D. I., Burt, D. M., Penton-Voak, I. S., Lee, K. J., Rowland, D. A., & Edwards, R. (1999). Symmetry and human facial attractiveness. Evolution and Human Behavior, 20(5), 295-307.
- Farkas, L. G. Anthropometry of the Head and Face. Raven Press. Standard reference for classical facial proportion canons, including facial thirds and fifths.