A score out of 100 built from four measurements of your face: how symmetrical it is, how its spacing compares to the proportions rated best in research, and how evenly it divides vertically and horizontally.
Choose one clear, front-on photo
Looking at the camera, even light, hair off your forehead, no
glasses.
JPG, PNG or WebP up to 10 MB, or drag one here
Your photo never leaves your device.
Geometry score
/100
Add a photo and this fills in.
Distinctive below 44
Varies 44 to 58
Mixed 58 to 72
Close 72 to 85
Very close 85 to 100
Symmetry
not yet /100
Averageness
not yet /100
Vertical balance
not yet /100
Horizontal balance
not yet /100
Where that sits
not yet
Typically 0 to 100 against this tool's four components
Taking the measurement
1
Take a front-on photo
Square to the camera, eyes level, even light, hair off your forehead and no glasses. Angle is the single biggest source of error in any of this.
2
Choose the file
The photo is read by your browser. It is not uploaded, not stored, and not sent anywhere.
3
Read the components, not just the total
The four parts are shown separately. They tell you far more than the combined number does, because they say which part of the geometry is doing what.
Example
Paired landmarks, measured against the midline.
Drawn from the landmarks our own measurement reads off this photo, not
placed by hand. It is a generated face, not a real person, and nothing
here is a verdict about it.
This measures four things from a photo of your face: how symmetrical it is, how close its spacing sits to the proportions rated best in the research, and how balanced it is vertically and horizontally. Those combine into a score out of 100. The formula and its weights are printed below.
How this is worked out
Symmetry
35%
Twenty paired landmarks compared across your face's midline, grouped into brows, eyes, nose, mouth and jaw. It carries the most weight because it is the best-evidenced component here: Rhodes (2006) reviewed the field and found symmetry attractive in both sexes and across cultures.
Averageness
35%
How close your eye-to-mouth and eye-to-eye spacing sit to the proportions Pallett and colleagues found rated best. Equal weight with symmetry because averageness replicates just as reliably, and because those proportions turned out to be the ones an average face already has.
Vertical balance
15%
How evenly your face divides into upper, middle and lower thirds. Low weight on purpose. Farkas measured 153 real faces against the classical canon and the vertical proportions were the worst-fitting of the nine.
Horizontal balance
15%
How close your face width sits to five eye-widths, the classical fifths canon. Low weight for the same reason as the thirds, and it is scored as distance from a convention rather than from an ideal.
Those weights are ours. They are a disclosed calibration for this
tool, not a formula the research agrees on, and no study anywhere
reports that a face works this way. What the number tells you is how
the geometry in your photo compares with the same geometry measured
the same way. It does not predict how any person will read your face.
What each measurement is
Reference bands are this tool's own, chosen for the landmarks it uses.
Where research reports a figure it is given separately, with the source,
because the two are different claims. And each one says what actually
moves it, which for most facial measurements is the more useful half.
A summary of this photo. Not a percentile, not a probability, not a validated attractiveness score.
What moves it Read the four components rather than the total, because they sort into two piles and only one is worth your time. Symmetry and spacing are structural - they do not move, and no routine, device or exercise changes them. What moves is everything the geometry does not measure: body fat, which changes facial definition within weeks; skin; hair; sleep; and the photograph itself. If you want one instruction from this page, it is that the lowest-scoring component is usually not the one to work on.
Symmetry /100
Formula
20 paired landmarks compared across the face midline
Research
symmetry is attractive in both sexes and across cultures - Rhodes (2006), a review plus meta-analyses
The heaviest weight, because it has the strongest evidence behind it.
Averageness /100
Formula
closeness of eye-to-mouth and pupil-to-pupil spacing to the values rated best
Research
faces rated best at eye-to-mouth 36% of face length and pupils 46% of face width - Pallett, Link & Lee (2010)
Those proportions turned out to be the ones an average face already has, which is why this carries equal weight with symmetry.
Vertical balance /100
Formula
closeness of the three facial thirds to an even split
Research
the equal-thirds canon fits real faces poorly - the vertical proportions were the worst of nine canons across 153 adults (Farkas et al., 1985)
Low weight for exactly that reason.
Horizontal balance /100
Formula
closeness of face width to five eye-widths
Research
the five-fifths canon held in about 37% of the faces Farkas measured (1985)
A distance from a drawing convention, not from an ideal.
This measures the geometry of your face from a photo and reports it as a score
out of 100. The formula is printed above. You can see every component, and you
can check the arithmetic.
That is the whole difference between this and the rest of the results page for
this search. Most tools in looksmaxxing return a number from a process they will
not describe, which makes the number unarguable rather than accurate.
What the score is measuring
Four things, weighted as shown above.
Symmetry compares twenty paired landmarks across your face’s midline, in
five groups. It carries the most weight because it has the best evidence behind
it. Rhodes reviewed the field in Annual Review of Psychology and found
symmetry, averageness and sexual dimorphism are all attractive, in both sexes,
across cultures.
Averageness measures how close your eye-to-mouth and eye-to-eye spacing sit
to the proportions Pallett and colleagues found rated best: about 36% and 46%.
It is weighted equally with symmetry because the effect replicates just as
reliably.
Vertical and horizontal balance measure how evenly your face divides into
thirds and fifths. Both carry low weight, and the reason is in the next section.
Why the golden ratio is not in the formula
Every competing tool weights it. Ours gives it nothing.
Our evidence page grades golden-ratio facial scoring as poorly
supported in the literature, and weighting a measure we publish a debunk of
would tell you plainly that the formula is decoration rather than reasoning.
There is a better version of that idea, and it is in the same research. Pallett
and colleagues ran four experiments looking for an ideal arrangement of facial
features. They found one: eye-to-mouth distance at roughly 36% of face length,
eyes roughly 46% of face width apart. Then they noted what those proportions
correspond to. They are the proportions of an average face.
So there is a measurable optimum in facial spacing. It replicates the averageness
finding from 1990. And it is not 1.618 and never was. Our
golden ratio calculator measures phi proximity anyway,
because people search for it, and then tells them this.
Why the thirds and fifths carry so little weight
The classical canons say a face divides into three equal vertical parts and five
equal eye-widths across. They are drawing conventions from the Renaissance.
In 1985, Farkas and colleagues measured 153 young adults against nine of those
canons. The two that held most often were both horizontal, at 40% and 37% of
faces. The vertical proportions fared worst of all nine. Their own conclusion
was that the canons do not represent average facial proportions, and reading
them as a prescription for an ideal has to be tested rather than assumed.
So a face that does not divide into even thirds has found the normal case. That
is why the component is worth 15 points and not 25, and why the
facial thirds tool leads with the same finding.
What this cannot tell you
It measures a still photograph. It does not measure you.
It says nothing about how people will respond to your face. The research
behind the weights describes averages across large groups of raters. An average
is a real thing, and it is not a prediction about one person. Expression,
grooming, health, voice, posture and manner are all absent from a still, and all
of them move how a face reads.
It is sensitive to your camera. Angle, lens and lighting all shift these
measurements. A phone held slightly off-square compresses one side of your face
and drops your symmetry score. If you want to compare two photos, take them the
same way.
It has no population to compare you against. Where you see a reference range
on this site, it comes from a published paper and is cited. We do not have a
database of measured faces, and inventing percentiles would be the easiest thing
on this page to fake.
What actually moves, if anything
Worth being straight about which components respond to effort.
Body fat changes facial definition, and it changes it fastest. It shows in
the jaw and cheekbones within weeks. The body fat estimate
gives you a figure to track.
Symmetry is largely structural. Some asymmetry is postural or dental and a
professional can address it. Most of it is bone, and it is worth knowing that
before spending money.
Spacing does not change. The distance between your eyes is fixed.
Anyone offering to move all four is selling something. The useful question is
not what your score is, it is which component is lowest and whether that one
responds to anything.
If this is making you feel worse
Measuring your face once out of curiosity is ordinary. Measuring it repeatedly
and feeling worse each time is not, and it is common enough in this community
that it is worth naming.
If that is where this is going, it matters more than any figure on this page.
Talk to someone: a doctor, or a mental health line in your country. That is not
a disclaimer, it is the honest advice.
This tool cannot answer that, and neither can any tool. What it can do is measure four geometric properties of your face and show you where they sit. Attractiveness is how people respond to you, which involves expression, grooming, health, voice and manner, none of which a still photo contains.
Is this an accurate attractiveness test?
It is accurate about geometry and silent about everything else. The measurements are repeatable: the same photo always returns the same numbers. What it does not do is predict how any person will react to your face, and no published formula does that.
How is the attractiveness score calculated?
Symmetry 35%, averageness 35%, vertical balance 15%, horizontal balance 15%. The full breakdown is printed on this page above, with the reasoning for each weight. You can see every component score alongside the total.
Why does this not use the golden ratio?
Because our own evidence page says golden-ratio facial scoring is poorly supported in the literature, and weighting a measure we publish a debunk of would tell you the formula is decoration. Pallett and colleagues found the best-rated proportions are the average ones, not phi.
Is this an AI attractiveness test?
A machine-learning model finds 478 landmarks on your face. The scoring after that is arithmetic you can check, not a neural network with an opinion. That distinction is the point: a published formula can be argued with, and a black box cannot.
Is my photo uploaded anywhere?
No. Everything runs in your browser. The page has no upload endpoint, and the site's content security policy blocks connections to anywhere but this domain, so you can verify the claim from the response headers rather than taking our word for it.
Does this work for men and women?
The geometry is the same either way. The research behind the weights covers both sexes: Rhodes found symmetry and averageness attractive in male and female faces. The Pallett proportions were measured on female faces, which is a limit worth knowing.
Why did my score change between two photos?
Camera angle, lens and lighting all move these measurements, and a phone held slightly off-square compresses one side of your face. Take both photos the same way before comparing them.
What is a good attractiveness score?
Above 85 means most components sit close to their reference bands; 75 to 84 is a close match with some differences; 65 to 74 is mixed. Those describe how this photo compares against the tool's own bands. They are not percentiles, and no study establishes a score anyone should reach.
Can I improve my score?
Some components move and some do not. Body fat changes facial definition and it changes it fastest. Symmetry is largely structural. Spacing is fixed. Anyone promising to move all four is selling something.
Does this measure my face card?
"Face card" is community shorthand for how strong a face reads overall. This measures geometry only, which is one input into that and not the whole of it.
Is this the same as a PSL score?
Same measurements, different output vocabulary. The PSL scale runs 1 to 8 and comes from the forums rather than from research. Our PSL test uses this engine and reports a range on that scale instead of a score out of 100.
Why is symmetry weighted highest?
Because it has the strongest evidence behind it. Rhodes (2006) is a review plus meta-analyses in Annual Review of Psychology, and it found symmetry attractive across cultures. The weights follow the evidence grading rather than the other way round.
Should I use this to choose a dating photo?
It is a poor tool for that. One recent study found improving a profile photo by one standard deviation moved match rates from 25% to 43%, and what moved them was things like expression, framing and clothing rather than facial geometry, which does not change between your photos.
Is caring about this unhealthy?
Wanting to look better is normal and does not need justifying. Measuring your face repeatedly, feeling worse each time, is a different thing. If that is where this is going, that matters more than any number on this page, and it is worth talking to someone about.
