Golden ratio face test: the real number, and what it is measured against

One ratio: face height divided by face width. Height runs from the top-of-forehead landmark to the bottom of the chin; width runs cheekbone to cheekbone. Both are distances between landmark coordinates, so the result does not depend on where you stood or how the photo was cropped. The raw figure is printed to three decimal places under your photo, next to your face shape, jaw angle and width-to-height ratio.

The 0-100 golden ratio score is the relative distance between your ratio and the target, scaled so that the median face lands in the middle of the range and a classically proportioned face lands at the top. It contributes 15% of the composite attractiveness score, the smallest of the four components, for the reason set out below.

The classical golden ratio face is measured from the hairline to the chin. A landmark mesh cannot see a hairline: its highest stable point is on the upper forehead, about 15% of face height lower. So a face that is exactly phi from hairline to chin measures around 1.38 in the window the mesh can see, and a face that measured 1.618 in that window would be unusually long.

When this test was first scored against 1.618 directly, the 95th percentile of real faces reached only about four fifths of the way to it and a top score was unreachable. The target is now phi mapped through the visible window, 1.38, calibrated on thousands of measured scans. This is the part most calculators leave out, and it is why their numbers cluster low or come with a suspiciously round percentage attached.

• Whether you are attractive. The research linking phi proportions to attractiveness ratings is weak; symmetry and averageness explain far more, and they explain only a modest share themselves.
• Anything about a face from outside the tradition that produced the ratio. The target is inherited from a European-descended aesthetic canon, and a lower score often means a different face shape, not a worse one.
• Depth. A single front-facing photo cannot see projection, and a chin-down or chin-up tilt changes the measured height by several percent. The result grades your head pose for exactly this reason.
• What to do about it. Nothing on this page suggests changing your face. The measurement is the product.

The method page has the full pipeline, the weights and every limitation.

What is a golden ratio face?

A face whose key proportions sit close to phi, 1.618, the ratio that recurs in classical art and architecture. The popular version compares face height to face width: a "golden ratio face" is one where height divided by width lands near 1.618. Beauty masks and most online calculators are built on that idea. On this test the ratio measured is exactly that - face height to face width - taken from landmark coordinates rather than a hand-drawn overlay.

How do I calculate the golden ratio of my face?

Divide face height by face width. Doing it by hand means guessing where the hairline and the edges of the face are, which is where most of the error comes from. This test places 478 landmarks on your photo, takes the distance from the top-of-forehead landmark to the chin as height and the distance between the cheekbone landmarks as width, and divides one by the other. The raw ratio is shown to three decimal places alongside a 0-100 score.

Why does my golden ratio say 1.3-something instead of 1.618?

Because the mesh cannot see your hairline. Its top landmark sits on the upper forehead, roughly 15% of face height below the hairline that the classical ratio is measured from. A face that is perfectly phi from hairline to chin measures about 1.38 in the window the mesh can see. That is the target the score is computed against, so a classically proportioned face scores highest instead of being penalised for the model's blind spot. Calculators that compare a landmark measurement directly to 1.618 give almost everyone a low number.

What is a good golden ratio face score?

The result shows where your golden ratio sits against 7,997 previously measured scans as an "ahead of N%" figure, attached from the 55th percentile upward and capped at 99%. A raw ratio close to 1.38 scores highest. Most faces land a little below it, which is another way of saying most faces are a touch wider than the classical ideal - and the classical ideal is one aesthetic tradition, not a law.

Is the golden ratio face scientifically proven?

No. Studies that test whether phi proportions predict attractiveness ratings find weak or inconsistent effects, and symmetry and averageness predict ratings considerably better. Faces rated highly by real people routinely miss the ratio, and faces that hit it are not reliably rated higher. That is why this test weights the golden ratio at 15% of the composite score - the smallest of the four components - and reports the raw number without dressing it up as a verdict.

Does the golden ratio face test upload my photo?

The free test never uploads your photo. The landmark model runs in your browser, no request carries the image, and there is no account or email step. The one flow on this site where a photo leaves the browser is buying the optional full report, which asks for explicit consent first.

• Face symmetry test: the heaviest component, measured alone
• The full facial attractiveness test, all six components
• Facial harmony: how proportion combines with symmetry
• The exact formula behind the score