Facial Thirds & Fifths

How this facial thirds works

You choose a clear, front-facing photo. The image is drawn into a canvas inside your browser, resized, and its metadata stripped — no copy is uploaded, stored or sent to a server. A small set of transparent statistics is computed from the pixels: a mirror difference between the left and right halves, the distribution of luminance and edge density across vertical and horizontal bands, and a rough projection of a skin-tone region. Those statistics are mapped onto six broad bands and combined with published weights into the Score.

We publish the whole formula at the methodology page because "formula shown" should mean something. The weights are D_sym 22%, D_gold 20%, D_thirds 18%, D_fifths 14%, D_eyes 14%, D_jaw 12%. They are our disclosed heuristic calibration — not a scientifically universal formula and not a prediction of how any person will rate you.

What your result means

Results are reported as a placement inside a broad reference band rather than a grade out of 100. A reading "inside the broad reference range" means the image statistic sat near the middle of the band we use; it is not a compliment, a diagnosis or a ranking. 100 is not the goal — balanced ranges are wide and strong-looking faces sit all over them.

What a single photo cannot support

A single front-facing photo cannot support side-profile claims. Where that is true, we say "not scored" instead of guessing. Nose projection, bridge profile, the true gonial angle and chin depth need a side view or a clinical measurement. Where a front photo cannot support a claim, we write "not scored" instead of inventing a number. This tool measures the photo, not your face directly. Lighting, angle, focal length and expression all move the reading. Lighting, head tilt, focal length, crop and expression usually move a reading more than the underlying geometry does.

Limitations, honestly

• This is not a 478-landmark clinical measurement. Estimated with a lightweight on-device geometry heuristic — not a 478-landmark clinical measurement. We say so next to every number, not in tiny footer text.
• Reference ranges carry bias. Aesthetic reference ranges in the literature carry documented bias. We treat them as rough bands, not ideals. We treat them as rough bands, never as ideals.
• The tool measures the photo, not your face directly. This tool measures the photo, not your face directly. Lighting, angle, focal length and expression all move the reading.
• No ranking, no percentiles, no predictions. We never rank you against an ethnicity-based ideal or against another person's face. The Score is a weak geometric signal. It does not predict matches, dates, replies, or your worth.
• Nothing medical or procedural. No procedures, surgery, fillers or injectables are recommended anywhere on this site.

Tips for a more consistent read

1. Use a recent, front-facing, well-lit solo photo with your whole face visible.
2. Face a window; avoid overhead light and harsh direct sun.
3. Keep the camera at eye level and at arm's length or further.
4. Remove sunglasses, hats and heavy filters; keep a neutral expression.
5. Retake in the same light and distance if you want to compare two readings.

Where the facial thirds reading fits

The facial thirds signal is one of six equally-labelled estimates. It is a weak signal about a photograph. It is not one of the things that decides whether someone replies to you. The parts of a dating profile with real leverage are the ones you can edit: your first photo, your bio, the app you choose and how you open a conversation. That is what the optional Dating Report is built around.

References and further reading

• Langlois, J. H., & Roggman, L. A. (1990). Attractive faces are only average. Psychological Science.
• Rhodes, G. (2006). The evolutionary psychology of facial beauty. Annual Review of Psychology.
• Willis, J., & Todorov, A. (2006). First impressions from faces. Psychological Science.
• Farkas, L. G. (1994). Anthropometry of the Head and Face (population-specific norms, cited here for their limitations).
• Lugaresi et al. (2019). MediaPipe: a framework for building perception pipelines — referenced only to state what this tool does not do.

The complete formula, weights and limitations are public: See exactly how the score is calculated. If we can't measure it honestly, we say so — before you pay.