Sample Full Face Report

Average.

Symmetry and averageness have no sub-measures: they are scored directly from the mirror mismatch (section 6) and the distance from the average face shape.

Oval, a clear match.

95% confidence in oval over the next closest shape

Jaw and chin, front view

Chin is narrow (28% of cheekbone width against 30% on the mean face), which lowers the score. The jaw and chin component scores 48.

Profile 60 of 100, above a typical face.

Chin sits 2.2% of face height in front of the vertical through the bridge of the nose, which scores well.

A 25° turn is not a profile photo. Depth comes from MediaPipe’s 3D estimate of the mesh, sharpened by the turned view, and the gonial angle is a soft-tissue angle that a beard hides. The nasolabial angle rests on a single mesh vertex; compare it with the mean face, not with clinical norms.

Neutral canthal tilt, −0.4°.

Inner and outer corners sit at about the same height, inside the ±2° band where the label can change from one scan to the next.

Canthal tilt is -0.4° (outer corners below the inner corners), flatter than the mean face's 1.7°, which lowers the score. The eyes and brows component scores 47.

59 out of 100: mild left–right differences.

Symmetry mirrors the subject’s right side onto the left across the fitted midline, over all 220 left–right point pairs of the mesh, after removing what a leftover head turn would explain. 100 is a perfect mirror. A typical face’s mismatch, about 3% of the eye-corner span, measures 74 on this scale. ±1.0 across the 8 front frames.

Left and right sides differ by 4.2% of the eye-corner span against about 3% on a typical face, most around the eyes, which lowers the score. The symmetry component scores 40.

Proportions 63 of 100, above a typical face.

Facial thirds are close to even (32/34/34%), which scores well.

The classical canon puts one eye width between the eyes (1.00). The score does not use this ratio; it scores the spacing of the eye centres against the mean face. The golden-ratio references are art-history canons, not research findings, and the score does not use them.

What would move the score.

Bone structure does not change with grooming. What changes is how it reads and how it measures: hair and beard sit on the outline points, posture and camera height change the angles the scan sees, and soft tissue covers the jaw and cheekbones. Each item answers one detractor from section 2.

1. 1Symmetry, 40 of 100

   Left and right sides differ by 4.2% of the eye-corner span against about 3% on a typical face, most around the eyes, which lowers the score.

   • Scan conditionsScan rested, with both eyes relaxed and open. A half-blink or a squint on one side counts as asymmetry.
   • Scan conditionsLight from straight ahead, a relaxed face and the phone square to it. Side light, a half-smile or one raised brow all measure as asymmetry.

2. 2Eyes and brows, 47 of 100

   Canthal tilt is -0.4° (outer corners below the inner corners), flatter than the mean face's 1.7°, which lowers the score.

   • GroomingCanthal tilt is set by bone and lid structure, and grooming does not move the measured corners. Brows with a lifted, tapered tail make the eye area read more upswept.
   • Posture and cameraPhone at eye level, head level and square to it. Pose correction removes most of a nod or a turn, not all of it.

3. 3Jaw and chin, 48 of 100

   Chin is narrow (28% of cheekbone width against 30% on the mean face), which lowers the score.

   • Hair and beardIf you grow facial hair, a short beard kept square at the chin broadens it.
   • Body compositionSoft tissue at the jaw corners rounds them. At a lower body-fat level the corners read more angular, and this measure follows the outline, not the bone.
   • Posture and cameraHead level, chin slightly forward, phone at eye level. A raised chin or a camera below the jaw softens the corners.

4. 4For an oval face

   • Hair and beardA textured crop or short quiff, which keeps your natural length and width as they are.
   • Hair and beardA classic taper or side part, short at the sides with a little more length on top.
   • GlassesMost frame shapes can work; frames about as wide as the widest part of your face tend to sit most naturally.
   • Hair and beardIf you grow facial hair, short stubble or a close-trimmed full beard follows your existing outline.

This report does not give medical or surgical advice.

How it was measured, and what it can’t see.

MediaPipe Face Landmarker tracked 478 points on 40 frames across 5 angles, on this device. Each angle’s burst was fused into one set of landmarks (the per-point median), the face was rotated to frontal in 3D, and the profile was read from the turned views. Every measure is an angle or a ratio, so image size and distance cancel out. The score is a weighted composite of symmetry ×0.20, averageness ×0.20, proportions ×0.15, eyes and brows ×0.15, jaw and chin ×0.15, profile ×0.15, mapped to 1–10 by a fixed curve.