PSL Calculator — Free Instant PSL Score on the 0–8 Scale

Free PSL calculator: upload a photo, get your 0–8 PSL score with tier name, percentile and all 8 metric readings in seconds. Every formula public, on-device, no sign-up.
Oct 12, 2026

A PSL calculator that actually calculates. Upload a photo, get your 0–8 PSL score, tier name, percentile and 8 metric readings — free, in seconds, with every formula on this page being the actual math, and the photo never leaving your device.

Seconds, not minutesMeasured on-device
No photo handy? Try an example:

🔒 Free measuring runs entirely on your device. When you unlock the AI deep report you can choose to attach your photo — used only to generate that one readout, then discarded, never stored. Scores are geometric measurements, not a judgement of your worth.

Algorithm, not magic

People ask whether a PSL calculator "uses AI" — usually meaning: can I trust the number? Here is the honest split: a small landmark-detection model finds 478 points on your face; everything after that is geometry you could verify with a ruler. No generative model looks at your photo and vibes a score. If you disagree with a reading, it is a disagreement about landmark placement — checkable, not mystical.

Three more things worth saying plainly:

  • The PSL scale is a community scale, not a clinical instrument. It comes from the PSL/lookism forums: 0–8, deliberately stricter than a casual 1–10, with the population median anchored at 4.0. No clinical literature validates it. It is a fixed ruler for comparing your face to itself over time — which is exactly why determinism matters more here than anywhere.
  • The same photo always scores the same. The engine is deterministic. Change the photo — angle, light, distance — and the geometry genuinely changes; that is measurement working, not drift.
  • It cannot see skin, style, haircut or expression. It reads geometry: angles, ratios, deviations from your facial midline. A great haircut moves your look more than most ratios ever will — the calculator just cannot see it.

How the calculator works

Four steps, fully deterministic:

  1. Detect — a landmark model places 478 points on your face (the same class of model used in medical imaging research).
  2. Measure — 8 metrics are read as raw geometry: angles, ratios, and deviations from your facial midline.
  3. Weight — the 8 sub-scores roll up into 5 weighted dimensions.
  4. Curve — the weighted overall maps onto the 0–8 PSL curve with the median anchored at 4.0.

The 8 metrics behind the score

MetricWhat it readsCommunity referenceTolerance
SymmetryMidline deviation across 8 paired landmarks< 1.2% of face width1.2%
Facial ThirdsUpper / mid / lower face height balanceEqual thirds±4.5%
Golden RatiosFace h/w 1.46 · mouth/nose 1.55 · eye/nose 1.0As listed0.12 / 0.25 / 0.14
Eye SpacingPupillary distance ÷ face width0.460.035
FWHRBizygomatic width ÷ nasion-to-upper-lip2.75*0.22
Jaw Width RatioBigonial ÷ bizygomatic width0.880.07
Canthal TiltOuter vs inner eye corner angle+6°±4.5°
Eye AspectEye height ÷ eye width0.330.06

*FWHR is calibrated in MediaPipe normalized coordinates, where it centers near 2.75 — not the academic photo-measurement value of ~1.85. Same geometry, different ruler. See the methodology.

Each metric scores 0–100 with a Gaussian falloff: 100 means dead-center on the reference band.

The five dimensions that weight them

DimensionWeightBuilt from
Symmetry20%Midline deviation
Proportions25%Facial thirds · golden ratios · eye spacing
Structure25%FWHR · jaw width
Eyes15%Canthal tilt · eye aspect
Harmony15%Consistency across all 8 sub-scores

Your score, your tier

The number lands on the community's tier map (male / female names for the same bands):

PSLTierWhere you sit
6.5 – 8Chad / StacyTop 0.1% — model-tier structure
6 – 6.5Chadlite / StacyliteTop 1% — very rare
5 – 6HTN / HTBClearly above average
3.5 – 5MTN / BeckyThe majority lives here
2 – 3.5LTN / LTBBelow average
0 – 2Very LowSevere disharmony

Two honest notes: the median is 4.0, not 5 — this scale is deliberately stricter than a casual 1–10, and a 5.5 already means clearly above average. And tier names are community vocabulary, not verdicts — the full breakdown lives in the tier list and the PSL scale guide.

Try it on an example face first

The calculator ships with example faces (AI-generated, spanning MTN / HTN tiers) — click one above the upload area and the full report runs instantly: score, tier, percentile, all 8 metrics, the radar. It is the fastest way to see exactly what you get before using your own photo. Example runs are free and never touch billing.

Same conditions, comparable scores

A PSL score is only comparable against another score taken the same way. The full test protocol in one paragraph:

  • Same angle — front-facing, chin level, no head roll.
  • Same distance & light — arm's length, even indoor light, no window behind you.
  • Neutral face — brows relaxed, mouth closed, no smile squint.
  • Same cadence — retake in 30 days under these conditions; the delta between two standardized runs is the real signal.

FAQ

The PSL cluster:

Narrower instruments, same engine:

References

Farkas LG. Anthropometry of the Head and Face. 2nd ed. New York: Raven Press; 1994 — the anthropometric datasets behind every reference band in the metrics table.

The PSL scale's origin: rating threads on the PSL/lookism forums (0–8, median anchored at 4.0) — a community convention with no clinical literature behind it, documented as such. Our methodology page publishes every anchor, tolerance and weight we map onto it.

Rhodes G. The evolutionary psychology of facial beauty. Annual Review of Psychology. 2006;57:199–226. doi:10.1146/annurev.psych.57.102904.190208 — averageness, symmetry and dimorphism carry attractiveness; no single number decides it.

Get your PSL score — one photo, ten seconds

Free, on-device, no sign-up — score, tier, percentile and all 8 metrics, every formula public.

Run the PSL calculator