Every measurement, target and weight behind our scores — plus what they can't tell you and what happens to your photo.
Last updated: 3 October 2026
Facemaxify scores faces by measuring proportions between facial landmarks and comparing each measurement with a published target. There is no hidden attractiveness model: every score on our free tools can be traced back to specific measurements, the target each one is compared with, and the weight it carries — all listed on this page.
Our approach rests on three principles:
The same photo always produces the same landmarks, measurements and scores.
| Score | Scale | Components and weights |
|---|---|---|
| Attractiveness test | 0–100 | Symmetry 35% · jawline 25% · youthfulness (philtrum) 20% · nose proportion 10% · eye spacing 10% |
| Face rating | 0–100 and 1–10 | Symmetry 30% · jawline 25% · golden ratio 20% · facial thirds 15% · nose proportion 10% |
| Facial harmony score | 0–100 | Symmetry 25% · jawline 20% · midface 15% · facial thirds 15% · nose 10% · lips 7.5% · eye spacing 7.5% |
| Looksmax score | 0–100 | Symmetry 25% · jawline 25% · canthal tilt 20% · facial thirds 15% · eye compactness 15% |
| PSL rating | 1–10 | Symmetry 25% · jawline 25% · canthal tilt 20% · eye shape 15% · facial thirds 15% |
| Measurement | How it's calculated | Target |
|---|---|---|
| Symmetry | Distance of 8 left–right feature pairs from the facial midline (nose bridge, nose tip and chin); smaller ÷ larger distance for each pair, averaged | Perfect balance (1.0) |
| Golden ratio | Mouth width ÷ nose width; nose-to-chin ÷ lower-lip-to-chin; outer eye distance ÷ mouth width | 1.618 each |
| Canthal tilt | Angle of the line from the inner to the outer eye corner, averaged over both eyes | Positive (+3.5° or more); highest around +5° to +9° |
| Hunter eyes | Eye openness (lid gap ÷ eye width, 55%) and tilt (45%) | Narrow, upward-tilted eyes |
| Eye spacing | Gap between the inner eye corners ÷ eye width; face width ÷ eye width | 1.0 and 5.0 (rule of fifths) |
| Facial thirds | Forehead, midface and lower face as a share of face height | 33.3% each |
| Midface ratio | Brows to nose base ÷ nose base to chin | 1.0 |
| Lower third | Philtrum, lips and chin as a share of the lower third | 22% / 44% / 34% |
| Philtrum ratio | Nose base to upper lip ÷ nose base to chin | 0.22 |
| Lip ratio | Upper lip height ÷ lower lip height (55%); mouth width ÷ gap between the eyes (45%) | 0.65 and 1.5 |
| Nose ratio | Nose width ÷ face width; nose width ÷ gap between the eyes | 0.275 and 1.0 |
| Chin ratio | Lower lip to chin ÷ lower third (55%); chin width ÷ jaw width (45%) | 0.45 and 0.50 |
| Jawline score | Jaw width ÷ cheekbone width (60%); angle at the chin between the jaw angles (40%) | 0.78 for the ratio |
| Cheekbone width | Cheekbone width ÷ jaw width (60%); cheekbone width ÷ face height (40%) | 1.2 and 0.75 |
| fWHR | Cheekbone width ÷ brow-to-upper-lip height | Scored highest around 2.0 |
| Forehead ratio | Forehead height ÷ face height (50%); temple width ÷ cheekbone width (50%) | 0.33 and 0.85 |
| Brow position | Gap between brow and upper eyelid ÷ eye width | 0.5 |
| Face shape | Face length ÷ width, forehead ÷ jaw width and chin angle, classified into 7 shapes | No score — a classification |
The targets are aesthetic reference proportions, not universal laws. They draw on classical proportion guides such as facial thirds and the rule of fifths, on golden-ratio guidelines, and on research into facial attractiveness:
A score describes the geometry of one photo. It isn't a measure of your worth.
The free tools don't use an AI model trained on people's attractiveness ratings, so they don't learn the taste or bias of a particular group of raters. They do rely on general proportion targets, which reflect a mostly Western aesthetic tradition. We state the targets openly so you can judge them, and each tool's guide notes where typical proportions vary.
Free tools: your photo is analysed in your browser and isn't uploaded to our servers.
Full Facemaxify analysis (signed in): your photo is uploaded and stored with your account so you can see your scan history. To generate the written reports, your photo and measurements are sent to third-party AI providers: Google's Gemini models produce the skin report and improvement plan, and OpenAI's image models produce the personal color and hairstyle analysis.
Face-rating tools can feed insecurity, especially when scores are treated as verdicts. We'd rather you use them as structured feedback: look at the breakdown, change one or two things that are within your control, and stop there. If you find yourself constantly checking scores, avoiding people, or fixating on features others wouldn't notice, talk to someone you trust or a mental health professional. Body dysmorphic disorder is common and treatable.
It places 478 landmarks on your face, measures ratios between them, compares each ratio with a published target and scores how close you are. Composite scores are weighted averages of those measurements, with the weights listed on this page.
The free tools use an AI landmark model to find facial points, but the scores themselves come from transparent geometry, not from a model trained on people's attractiveness ratings.
Not with the free tools — they run in your browser. The signed-in full analysis uploads your photo to store your scan history and sends it to Google and OpenAI models to generate its written and visual reports.
The same photo always produces the same score. Accuracy depends mostly on the photo: a straight-on, neutral, well-lit picture taken from arm's length or further gives the most reliable result.
Yes. The free tools use the same general proportion targets for everyone, and their guides note where typical proportions differ between people and populations.
Only partly. Scores measure facial geometry in one photo. Skin, expression, style and personality — which strongly affect real-world attractiveness — aren't measured.