How the scoring works

30% · Symmetry

The symmetry dimension uses paired visible mesh points and a roll-corrected mirror-error approximation. Matching left and right in a photo increases the reference score. This is a two-dimensional estimate; lighting, expression and perspective can change the detected outline. It does not establish anatomical symmetry.

25% · Visible thirds

We divide the visible landmark mesh into three vertical spans. The mesh top is not a detected hairline, so these are visible-landmark thirds, not a clinical measurement of traditional facial thirds. Their deviation from equal spans is mapped through the product scoring curve.

25% · Facial fifths

Five horizontal spans are measured between the face edges and inner and outer eye corners after eye-line roll correction. Each span is compared with one fifth of the measured width. The score summarizes the deviation from that chosen reference, rather than asserting that equal spans are necessary for an attractive face.

20% · Reference ratio

The visible mesh length-to-cheek-width ratio is compared with 1.618 as one openly chosen product reference. A smaller difference increases this component. This is not a complete golden-ratio mask and does not establish that any one face proportion is universally ideal.

What the result can support

Each dimension has a fixed weight: 30%, 25%, 25%, 20%. Missing dimensions do not get a replacement score or a silent redistribution of their weight. The complete tool page shows the raw values, score curve and version alongside the result.

Rejected photos receive no overall score. A caution result is explicitly provisional. We do not infer medical conditions, identity, personality, or a percentile among people from these measurements. A lower score is a mismatch with a reference in this photo, not a judgment about you.