Golden ratio face calculator

By the LooksLab Team · Updated September 24, 2026

Your golden ratio result in 30 seconds

The calculator above places 478 landmarks on your face, levels your head to the midline, and measures five ratios that the golden ratio canon says should equal 1.618. For each one you get your value, how far it sits from 1.618 and a match score. The average of the five is your golden ratio match out of 100.

The photo is measured in your browser and is never uploaded. The scores are app estimates of geometry. They are not an objective measure of your worth or a medical finding.

Two things make this result different. First, we only score ratios that landmarks can measure: no hidden hairline guesses. Second, we show the proportions that research found to matter more than 1.618, so you can see both sides.

What is the golden ratio?

The golden ratio, written φ (phi), is about 1.618. Euclid defined it as dividing a line so that the whole is to the longer part as the longer part is to the shorter part. Solve that and you get (1 + √5) ÷ 2 = 1.6180339…

In the 1850s the German writer Adolf Zeising called the golden section a universal law of beauty, fully realized in the human body. Many claims that followed do not survive measurement. A 2024 review in Maxillofacial Plastic and Reconstructive Surgery notes that measurements do not support the famous Parthenon claim, and that there is no sign Leonardo da Vinci used the golden ratio in his drawings of human proportions (Naini 2024).

In faces, the idea became popular through the orthodontist Robert Ricketts, who proposed in 1982 that facial beauty could be measured with phi (Ricketts 1982), and through Stephen Marquardt's "phi mask", a face template built from golden proportions (Marquardt 2002).

The five ratios this calculator measures

We use the frontal points from Ricketts' method as applied in a 2016 study of attractive adults (Khan et al. 2016): the outer eye corners, the outer edge of the nose, the mouth corners and the soft-tissue chin. The vertical gaps between those four levels are a (eyes to nose), b (nose to mouth) and c (mouth to chin).

That study selected 130 attractive North Indian adults (65 women, 65 men) with a rating panel. Five of the women's seven height ratios landed near 1.618, but only two of the men's did (both of them hairline ratios), and the width ratios missed 1.618 in both groups. The width values were closer to 1.414, the so-called silver ratio. All five ratios count toward your score, and any of the five can be named as your biggest opportunity, as explained below.

How each point is found:

• Eyes: the outer eye corners (MediaPipe landmarks 33 and 263).
• Nose: the widest point of the nostril wings (64 and 294). The same points give the nose width.
• Mouth: the mouth corners (61 and 291).
• Chin: the lowest point of the chin on the midline (152).

The calculator first rotates the landmarks so the line from the forehead (10) to the chin (152) is vertical. That removes head tilt, so a leaning head does not change the vertical gaps.

How to read your score

Each ratio gets a match score: 100 at exactly 1.618, minus 2 points for every 1% it sits away. A ratio 5% above 1.618 scores 90; one 20% below scores 60. Your golden ratio match is the average of the five.

Each ratio is also colored. Green means within 5% of 1.618, blue within 12%, and orange further out. Orange is a description, not a flaw: the mouth ÷ nose width ratio is orange for many faces, including both attractive groups above. A face with exactly the average ratios of the attractive women in that study would score about 81: 99, 96 and 91 on the height ratios, but 63 and 55 on the width ratios.

Keep one distinction in mind. The match score measures closeness to one number. It does not measure attractiveness. Because 1.618 explains little of how attractive faces are rated, a face can score 70 and still be rated attractive, and a face can score 90 without standing out. For a result built on the factors with better evidence, take the attractiveness test.

How your biggest opportunity is chosen

Under the score, the result names one ratio as your biggest opportunity: the one of the five that sits furthest from 1.618, whether it is a height ratio or a width ratio. The title gives the ratio and your value, such as "Mouth ÷ nose width: 1.31 vs 1.618", and the text says which gap is long or short compared with the canon and what can and cannot change. The same ratio is marked "Furthest" in the ratio list.

Being 10% above 1.618 counts the same as being 10% below it. That is how the canon works, but research gives some directions a clear meaning, and the result says so:

• Mouth ÷ nose width below 1.618 is the usual pattern. The attractive adults in Khan et al. averaged 1.32 (women) and 1.21 (men), so on most faces this is the ratio furthest from 1.618.
• Eye span ÷ mouth width above 1.618 is also the usual pattern: the same groups averaged 1.98 and 1.90.
• A longer lower face is more common in men. In a 3D study of 272 Turkish and Japanese adults, the women in both groups had a shorter lower face than the men (Tanikawa et al. 2021). A lower-face ratio off 1.618 in that direction is often a sex difference, not a flaw.
• Pupil gap ÷ face width is compared with the 46% average described below, and the result says whether your eyes are wider-set or closer-set. In that study ratings peaked at the average, so neither direction is favored.

Why face length ÷ width is not scored

The best-known golden ratio claim is that a beautiful face is 1.618 times as long as it is wide, measured from the hairline to the chin. That ratio needs the hairline. The face mesh used here has no hairline point, and hair often hides the hairline anyway. The top landmark of the mesh (10) sits on the upper forehead, below the hairline.

So the calculator shows face length ÷ width with a plus sign, such as 1.18+. Your real hairline-to-chin ratio is higher than that number by an amount a photo cannot tell us. Any tool that scores this ratio against 1.618 from landmarks is either guessing where your hairline is or comparing a shorter length with a target that assumes a longer one.

If face length and width are what you care about, the face shape detector reads the same widths as a shape (oval, round, square, heart or oblong), which is more useful than a single ratio.

What research says about the golden ratio and beauty

The golden ratio has been tested directly, and the results are weak.

• Few golden proportions predict ratings. Kiekens and colleagues had 76 lay raters score photos of 64 adolescents, then measured 19 golden proportions from Ricketts' method. Only 4 of the 19 related to the ratings, and together they explained 16% of the variance, which the authors called too small to be of clinical importance (Kiekens et al. 2008).
• The phi mask fits a narrow face type. Holland found that Marquardt's mask best describes the proportions of masculinized white women as seen in fashion models, is poorly suited to many non-European faces, and uses a faulty method to judge fit (Holland 2008).
• Some studies find a link. Pancherz and colleagues measured cover models and former orthodontic patients and found that attractive faces deviated less from the divine values than non-attractive ones (Pancherz et al. 2010). The evidence is mixed, not zero.
• The current verdict. A 2024 review concluded that there is no convincing evidence that the golden ratio is linked to idealized human proportions or facial beauty, and no evidence to support its use in facial surgery planning (Naini 2024).

The "new golden ratios": what matters more

In 2010, Pallett, Link and Lee took a different approach. They edited photos of women's faces so that only the spacing of the eyes and mouth changed, and asked people to compare them. For each face, attractiveness peaked when the vertical distance between the eyes and the mouth was about 36% of face length (hairline to chin), and the distance between the pupils was about 46% of face width (between the inner edges of the ears). These values matched the average of 40 faces, and they were significantly different from the golden ratio values (Pallett, Link & Lee 2010).

The calculator shows your version of the second ratio under the result: pupil distance ÷ face width, against the 46% average. It is an approximation, because we measure face width at the face outline at eye level, not between the ears. The 36% ratio needs the hairline, so we do not measure it. The study used only white women's faces, and the authors note that the ideal may differ for men and other groups.

The broader research points the same way. A review with meta-analyses found that averageness, symmetry and sexual dimorphism are attractive in both men's and women's faces and across cultures (Rhodes 2006). To check those directly, try the face symmetry test or see where your geometry lands on the PSL scale.

Limits of any photo measurement

Photo ratios are sensitive. These are the main sources of error:

• The hairline. It is not visible to landmarks, which is why face length is not scored.
• Lens distortion. A phone held close exaggerates what is nearest to it. A model from Rutgers and Stanford researchers estimated that a selfie taken about 12 inches away makes the base of the nose look about 30% wider than a photo taken from 5 feet (Ward et al. 2018; Rutgers summary). A wider nose lowers your mouth ÷ nose ratio.
• Pose. Head tilt is corrected, but turning left or right and tipping the chin up or down change the gaps. The calculator warns you when your head is turned.
• Expression. A smile or open mouth moves the mouth corners. The result adds a note when it detects one.
• Short segments. The nose-to-mouth gap (b) is short, so a shift of one or two pixels in a small photo can move a ratio that uses b by several percent. Take two photos in the same setup and compare before you read much into a single number.

What you can and cannot change

No exercise or skincare routine moves your eye, nose and mouth levels. What changes is how your proportions read in person and in photos:

• Outline. Hair and facial hair change where the face seems to start and end. Pallett and colleagues suggested this is one reason a hairstyle can make the same face look more or less attractive: it changes the perceived face length and width.
• Lower face. Body fat changes how defined the jaw and cheek edges look, and with them how long the lower face seems.
• Eye area. Brow shape and under-eye care change how the eye region reads. The canthal tilt test measures the angle of your eye corners.
• Photos. Camera distance and height change the measured ratios themselves, as shown above.

Treat an orange ratio as a styling cue, not a verdict. The LooksLab app turns measurements like these into a routine for skin, jawline, hair, eye area and more.

How to take a good photo for this test

Use a front photo with your head level, your face relaxed and your mouth closed. Hold the camera at eye level and at least an arm's length away, or ask someone to take it from further back. Pull your hair off your face so the outline is visible, and use soft, even light. Our guide to taking photos for face analysis covers each step.