PSL scale test

Updated 15 October 2026How we measure

What your PSL number means

Your result is a number from 1 to 8 with one decimal, and a band. It is our measured attractiveness score, re-expressed on the range the PSL scale uses. A typical face lands at 4.5, the middle of the range. The band says in plain words where your number sits; the bands have the same edges as the tiers of our attractiveness test, so the same face lands in the matching band on both pages.

Two things to keep in mind when you read it. First, the decimal is less certain than it looks: light, expression and a slightly turned head can move the underlying score by a few tenths, so the band is the more stable part of the result. Second, in practice the formula runs from about 1.2 to about 7.5 on this scale. A face with perfect symmetry and every measurement at its ideal still stops short of 8, which matches how the scale is usually described: 8 is a ceiling, not a score.

How we calculate it

There is no separate PSL model. The scan measures your face with Google’s MediaPipe Face Landmarker, 478 points in your browser, and computes the same documented attractiveness score as our attractiveness test. That score combines six components, each scored from 0 to 100 where a typical face scores 50:

• Symmetry (20%): how closely the two sides of the face mirror each other.
• Averageness (20%): how close the face’s overall shape is to the average face.
• Proportions (15%): how even the thirds and fifths are, and the eye spacing.
• Eyes (15%): canthal tilt, how open the eyes are and their size for the face.
• Jaw and chin (15%): jaw and chin width against the cheekbones, and the jaw angle.
• Profile (15%): jaw angle, chin and nose projection from the side views of the scan. A single photo has no profile, so the other weights are rescaled without it.

The weighted total becomes a score from 1.0 to 10.0, where a typical face scores 5.0. That score is then placed on the 1–8 range with a fixed curve: straight lines through four points.

The knots are chosen for two reasons. The typical face sits exactly in the middle of the range, and the edges of the attractiveness tiers (4.0, 5.5, 7.0 and 8.5) land on round PSL numbers (3.5, 5.0, 6.0 and 7.0), so the bands on both pages always agree. The curve rises everywhere, so a better-measuring face never gets a lower PSL. Every weight and threshold behind the score is on the methodology page.

A worked example

Say the scan scores your face 6.2 on the attractiveness scale. That sits between the knots at 5.5 and 10.0, where each point of score is worth two thirds of a PSL point, so the estimate is 5.0 + (6.2 − 5.5) × ⅔ = 5.47, shown as 5.5: “Above the middle”, the same band as the attractiveness tier “Above average”. A score of 4.6 sits on the steeper middle stretch, where a point of score is a point of PSL: 3.5 + 0.6 = 4.1, in the middle band. A score of 3.8 gives 3.3, the lower band.

The steeper middle stretch is deliberate: faces near typical score between about 4 and 6, and spreading that range a little keeps differences there visible, while the long, flatter stretch above 5.5 keeps the top of the scale hard to reach, as the forum scale intends.

Photo or scan: why your number can differ

You can get a PSL estimate from a single photo, but it is not the same measurement as the scan, for three reasons:

• No profile. A front-facing photo has no side view, so the profile component (15%) is left out and the other five are re-weighted to fill the gap. A strong or weak profile can move the scan’s result either way.
• One frame instead of eight. The scan takes the median of eight frames for every point, which removes blinks and jitter. A photo is a single frame, so its noise stays in.
• The lens. Photos are often selfies taken close to the face, which enlarge the nose and narrow the face. The scan asks you to hold the phone at arm’s length.

If a photo and a scan disagree by more than a few tenths, trust the scan, and if two scans disagree, the conditions are still moving the result: even out the light and try again.

Can your PSL change?

The bones that set most of the geometry don’t change in adults without treatment, so the estimate is fairly stable from scan to scan. What does move it:

• Facial fat. The amount of soft tissue over the jaw and cheekbones changes their definition, which the jaw and profile components see. Research by Coetzee and colleagues found that the perceived weight of a face predicts how healthy and how attractive it is judged to be.
• How you hold your face. A relaxed, neutral expression, chin level and the camera at eye level give the most accurate reading; squinting or a tilted head lowers it.
• Age. Over decades the soft tissue of the face changes, and so do the measurements.

Grooming, skin and hair change how you look to other people, but not this number, because the scan only measures geometry. Forum claims that exercises reshape the bones of the face have little evidence behind them; our guide to mewing before and after goes through what has actually been studied. Questions about jaw or bite treatment belong with a dentist or orthodontist; this page is not medical advice.

What PSL stands for, and where the scale comes from

PSL is made of the initials of three now-defunct online forums where the scale was popularised in the 2010s. We don’t spell the names out here, because the names themselves are not fit for print. All three were communities that discussed looks and dating, often in harsh terms. The scale they spread rates a face from 1 to 8, with the middle numbers standing for the average range and the top numbers treated as rare. It has no scientific definition and no agreed method: one person’s 5 is another’s 4.

The forums also attach a label to each number. Most of those labels are insults, and we don’t use them on this site. The numbers are useful shorthand that many people already know, so this page keeps the numbers, gives each band a neutral description, and leaves the labels out.

If you came here looking for the Pakistan Super League, this isn’t it: PSL on this page is the rating scale.

Why most faces sit in the middle

The score is an average of six components, and each component is itself built from several measurements. A face is rarely far from typical on all of them at once: a strong jaw and a short lower third, for example, pull in different directions. Averaging many measurements pulls most results towards the middle, which is why most faces land in the middle band and why the top band is hard to reach.

The formula also builds in a finding from attractiveness research: faces close to the population average are rated attractive. Composites made by averaging many faces are rated more attractive than most of the individual faces in them, a result first shown by Langlois and Roggman in 1990 and confirmed many times since. So the average face itself scores well above the middle here, about 7.4 on the attractiveness score from a full scan, while a typical individual face, with its own small asymmetries and departures from the average, scores 5.0.

What the number can’t see

The PSL scale on forums is usually applied to photos, with everything a photo shows: skin, hair, body fat, grooming, style, expression and the photo itself. Our estimate sees the geometry of your face and nothing else. That makes it reproducible, but it also means a large part of what people respond to is missing:

• Skin, hair and grooming. Skin, a haircut that suits the face shape and a well-kept beard change how a face looks to other people without changing its geometry.
• Expression. The scan asks for a neutral face because expression moves the landmarks. In real life, people see you smiling and talking.
• Taste. Raters agree about faces more than the saying “beauty is in the eye of the beholder” suggests: a large meta-analysis found strong agreement within and across cultures. But Hönekopp showed that private taste is about as strong as shared taste, and a twin study by Germine and colleagues found individual face preferences are shaped mostly by each person’s own experience, not genes. No single number can capture what one particular person finds attractive.
• The camera. A close phone camera enlarges the nose and narrows the face, and one-sided light hides points on half of it. MediaPipe’s model card reports a mean landmark error of about 2.6–3.2% of the distance between the eye centres, which is why the scan fuses eight frames.

Reading the number without letting it read you

A PSL number is a rough description of shape. It is most useful for one thing: seeing which parts of your face measure close to typical and which don’t, so you know what a haircut, a beard, glasses or a different camera angle could change. It is least useful as a ranking of you as a person, which is how the forums use it.

Practical points that move the result for real reasons:

• Scan in even light, facing a window, with the phone at arm’s length and eye level.
• Keep a relaxed, neutral face and your chin level.
• Pull hair off the forehead and take glasses off, so every point the formula needs is visible.
• Use the full scan rather than a photo: the profile component needs the side views.

Adults only

Like the attractiveness test, the PSL test is for people aged 18 or older, and the scan asks you to confirm your age before it opens the camera. Faces keep changing through the teenage years, the scores are not meant for growing faces, and rating the looks of minors is not something this site does.

The free result and the full report

The free result is your PSL estimate, its band and the attractiveness score it came from. The Full Face Report, $14.99 once, shows that score broken into its six components with your strongest and weakest features in plain sentences, your profile from the side angles, and what would move your score for your face: grooming, hair, beard, posture and camera height. It is built on your device from the same scan.

For the individual measurements behind the number, try the face symmetry test, the golden ratio face test or the facial ratios calculator.

Related reading

• Hunter eyes, explained
• Facial symmetry explained
• The golden ratio in faces: what the evidence says

Last updated 15 October 2026.