Free AI face rater and score. Symmetry and proportions — plus separate jawline and face-shape measurements — from 478 facial landmarks, rated 0–100 with the formula shown in your browser.
Upload a photo
Drag and drop or choose a file
JPG, PNG up to 10MB
Your photo never leaves your device
A reproducible way to test facial harmony from a photo is to measure visible geometry instead of asking strangers or relying on an unexplained rating. This test maps 478 landmarks, scores bilateral symmetry and proportional balance from 0–100, and shows the formula. The photo processing runs in your browser: your photo is not uploaded for the free test, while derived scores are saved temporarily so your result link and optional report can work.
Treat the number as a private baseline, not a verdict — the point is knowing which measurable features drive your score, because those are the ones you can actually work on.
Higher scores mean the photo's landmarks sit closer to this tool's selected symmetry and proportion targets. The bands below come from the same current thresholds used by the live result, so the explanation cannot disagree with the tier you receive:
87–100 — Showstopper
Top 5% — ahead of 95% of Face Report scans. Closest alignment across this tool's four geometry components.
83–86 — Standout
Top 15% — ahead of 85% of Face Report scans. Strong measured harmony, with only modest variation across components.
78–82 — Glow Up
Top 35% — ahead of 65% of Face Report scans. Above-average measured balance with identifiable areas of variation.
60–77 — Rising
Solid foundation, clear room to push higher. A mixed geometry profile where photo conditions and individual components matter more than the headline number.
Below 60 — Foundation
Distinctive features — biggest upside potential. Further from this tool's selected geometric targets. Distinctive faces and expressions can fall here without being less attractive to people.
Your score is a weighted sum of four geometric components. The components are informed by facial-aesthetics research, while the exact weights are this tool's disclosed production calibration—not a scientifically universal formula or a prediction of how any person will rate you.
How closely paired landmarks on the left and right sides mirror across a fitted facial midline. Bilateral symmetry is widely studied in perceived attractiveness; 35% is this tool's disclosed product-formula weight, not a universal scientific constant.
How evenly the visible landmark window divides into upper, middle, and lower sections. The target is adjusted for MediaPipe's upper-forehead landmark rather than pretending the model can see the true hairline.
How the face divides horizontally into five eye-width segments — ear to outer eye, outer to inner eye (one eye width), inner-eye gap (one eye width), and so on across to the other ear. Fifths balance is a centuries-old canon for upper-face proportion.
How closely your face length-to-width ratio matches phi (1.618), adjusted for the visible landmark window. It has the lowest weight because one length-to-width target captures less geometry than the bilateral and multi-section balance measures.
Several decades of research have converged on a few well-supported findings about what predicts perceived attractiveness across observers and cultures.
What's well-supported
What an AI photo test misses
Be honest with yourself about what this tool can and can't do
Front-facing, eye-level camera
Tilted angles distort vertical proportions and inflate apparent asymmetry. Hold the camera at eye level and roughly arm's length away.
Neutral expression
A relaxed, neutral expression gives the most stable measurement. Smiles dramatically alter mouth, cheek, and jaw landmarks and tend to lower the symmetry component.
Hair pulled back
Hair covering the hairline or jaw outline blocks landmark detection. Tie hair back so the full face contour is visible.
Avoid wide-angle selfies
Phone selfie cameras distort proportions, especially close to the lens. Use a back camera at arm's length, or have someone else take the photo from a few feet away.
Even, soft lighting
Strong shadows shift apparent feature edges and exaggerate asymmetry. Diffused front lighting (window light works well) gives the cleanest measurement.
Yes — with the formula shown. It combines four geometric components (symmetry, facial thirds, facial fifths, and golden-ratio proximity) into a reproducible 0–100 facial-harmony proxy. It does not measure every quality that shapes real-world attraction, and the disclosed weights are product calibration rather than a universal scientific formula.
Upload (or take) one front-facing photo with your face level and hair off your forehead. The test detects 478 facial landmarks in your browser, measures symmetry and proportions, and returns a 0–100 score, tier, and one personalized measurement preview. Your photo is never uploaded for the free test; derived scores are saved temporarily so a result link and optional paid report can work. No login is required.
On the current calibrated scale, 78+ indicates above-average measured balance, 83+ indicates a strong geometry profile, and 87+ is the highest tier. These are bands within this tool, not universal beauty percentiles or a verdict on how people see you.
Two key differences: the formula is disclosed — symmetry 35% + thirds 25% + fifths 25% + golden ratio 15% — and the measurements come from MediaPipe's 478 facial landmarks in your browser. That makes the same photo reproducible under the same engine version while keeping the limitations visible.
You can make the measurement more consistent with a straight-on camera, neutral expression, even lighting, and a non-wide-angle lens. Hairstyle, grooming, and makeup can change how proportions read visually, but facial exercises are not a reliable way to change adult skeletal symmetry. Treat repeated scores as photo-condition baselines, not medical or anatomical targets.
Three common reasons. First, photo conditions: tilted angles, smiles, asymmetric expressions, and poor lighting all reduce scores significantly. Try a neutral, front-facing photo with even lighting. Second, this scale is strict near the top: 90+ requires close alignment across all four measured components. Third, the score measures distance from selected geometric targets, which do not capture every form of attractiveness.
The aesthetic-ideal ranges used here come from cosmetic surgery literature and Western anthropometric studies, which have well-documented Eurocentric bias. The tool doesn't classify by race or gender, but the ratio targets reflect a particular aesthetic tradition. Faces from non-European populations may score lower not because they're less attractive but because their natural proportions deviate from these specific norms. Treat the score as a measurement against one tradition, not a universal judgment.
It can be a useful tiebreaker if you have several similar photos and want to pick the one with the best lighting and angle (which strongly affect the score). But research on dating-app photo selection consistently finds that warmth, approachability, and a genuine smile predict matches better than mathematical harmony. A photo with a slightly lower harmony score but a genuine smile usually outperforms a higher-harmony but neutral one.
Decades of research (Rhodes, 2006; Little, 2014; and others) consistently identify several factors that correlate with attractiveness ratings: bilateral symmetry, averageness (faces close to the population mean), sexual dimorphism (gender-typical features), and clear skin. Beyond these, expression, behavior, vocal qualities, and cultural context strongly modulate perceived attractiveness. The harmony score captures the symmetry + proportional aspects; the others require a real human, not a photo.