How the golden ratio test works
- 1
Add a front-facing photo
We check the angle, light and that there is one face. Proportions change with head tilt, so a level photo matters.
- 2
478 points on your device
A face model in your browser finds the outline of your face, eyes, nose, mouth and chin. The photo is not uploaded.
- 3
Ten ratios against 1.618
Each ratio is compared with the golden ratio. The score averages how far they are from it: 100 would mean every ratio equals 1.618 exactly.
The ten ratios
These are the ten ratios our tool compares with 1.618. Widths use left-right points, and heights are measured along the vertical centre line, which removes in-plane head roll.
Face length to cheekbone width
DeviceThis divides the distance from the top of the face outline (mid-forehead, not the hairline) to the bottom of the chin by the widest span across the cheekbones. Charts that start at the hairline give a longer face length, so this ratio can differ from a tape-measure chart.
Mouth width to nose width
DeviceThis compares the horizontal distance from one corner of the mouth to the other against the width across the nostrils. A wider smile changes this ratio directly, so a relaxed, closed mouth gives the clearest reading.
Outer eye span to mouth width
DeviceThis divides the distance from the outer corner of one eye to the outer corner of the other by the corner-to-corner width of your mouth. It checks whether your eye span is about 1.618 times your mouth width.
Pupil distance to nose width
DeviceThis divides the horizontal distance between the centres of your pupils by the width across your nostrils. Photos taken very close to the face can distort relative widths, so hold the phone at arm's length.
Cheekbone width to mouth width, and to jaw width
DeviceThese two ratios divide your widest cheekbone span by your mouth width and by your jaw width. Together they compare the width of the midface with the width of the mouth and the jaw.
Four vertical height ratios
DeviceThese four measurements compare vertical segments along the face midline: total face height divided by brow-to-chin height, brow-to-chin divided by nose-to-chin, nose-to-chin divided by mouth-to-chin, and nose-to-mouth divided by mouth-to-chin. Tilting your head up or down can change these vertical distances and shift all four ratios.
What the golden ratio is and how face charts use it
The golden ratio, written as Phi and equal to about 1.618034, describes a proportion where the whole length divided by the larger part equals the larger part divided by the smaller part. Many online face charts apply this number to pairs of facial distances, asking whether one feature is roughly 1.618 times wider or taller than another. Our Golden ratio match score checks ten such pairs on your photo so you can see each ratio alongside its measurement lines.
The golden ratio is different from other proportion rules, such as equal facial thirds or eye spacing. Those rules test other relationships, whereas a golden ratio test looks specifically for 1.618 across pairs such as cheekbone width to jaw width and nose-to-chin to mouth-to-chin. Each of the ten ratios is shown separately, so you can see which ones are close to 1.618 and which are not.
a/b = (a+b)/a ≈ 1.618
We draw the measurement lines directly on your photo so you can check where each point sits on your forehead, eyes, nose, mouth and chin. If you want to share your result, you can save a summary card, and your photo is off by default. Because the tool runs in your browser tab, you can try different lighting or camera distances without sending images to a server.
How our tool measures the ten ratios step by step
When you select a photo, the page first runs a quality check before calculating any numbers. It rejects images with no face, more than one face, a face that is too small or too dark, or a head turned or tilted too far. Passing this check helps the ratios reflect your features rather than a steep camera angle.
Next, MediaPipe Face Landmarker runs inside your browser tab and estimates 478 three-dimensional points across your face. It builds on research into real-time 3D face mesh estimation from a single camera on mobile GPUs. The model file is about 4 MB and downloads once, and your photo stays on your device. The same photo always gives the same numbers.1

Pink dots mark detected facial features; pink lines connect the measured widths, face length and eye corners.
From those points, the engine computes five width ratios, one face length to cheekbone width ratio and four height ratios, with heights measured along the midline, which removes in-plane head roll. For each of the ten ratios, it finds how far your ratio sits from 1.618034 as a fraction of 1.618034, averages those ten relative differences, subtracts that average from 1, and multiplies by 100. The result is clamped between 0 and 100. For example, an average difference of 24.4% gives a score of 75.6, and an average difference of 20.3% gives 79.7.
What the research says about golden ratios and beauty
The studies below do not support 1.618 as a universal rule for facial attractiveness. Pallett, Link, and Lee (2010) tested whether an ideal feature arrangement optimizes attractiveness by comparing faces with identical features but different eye-to-mouth and eye-to-eye distances. They found that attractiveness was optimized when the vertical distance between the eyes and mouth was about 36% of face length and the horizontal distance between the eyes was about 46% of face width. The authors report that these “new” golden ratios match those of an average face.2
Other researchers have examined the “Phi mask” that Stephen Marquardt derived from the golden ratio and presented as an ideal facial archetype. Holland (2008) reported that the method used to test how well faces fit the mask is faulty. He concluded that the mask does not appear to describe an ideal face shape, because its proportions are inconsistent with what most people prefer.3
Earlier anthropometric work also found wide natural variation. Farkas and colleagues (1985) tested nine neoclassical proportion formulas, which are older face-proportion rules and not specifically golden ratio rules, in 153 young adults. They found the vertical proportions matched worst, reported large variability in feature sizes across normal faces, and concluded that the canons do not represent average proportions and that using them as a prescription for ideal proportions must be tested.4
How to read your score against our calibration band
Our face model fits a standard mesh to each photo, so scores tend to fall in a fairly narrow band rather than spanning from 0 to 100. Our reference band comes from 133 real front-facing photos from Pexels, used as a technical mesh reference and not as a representative sample of any human population. On these calibration faces the middle 80% scored between 70.7 and 79.4, with a median of 75.8. A score in the mid-70s means your result sits near the middle of that calibration range.
Look at the individual ratio breakdown below your main score to see what pulled your number up or down. For instance, you might see your cheekbone width to jaw width ratio sitting close to 1.618 while your nose-to-mouth to mouth-to-chin ratio sits farther away. One or two ratios far from 1.618 can pull down the ten-ratio average even when your other ratios are close.
Number of photos
- Median:
- 75.8
- Middle 80%:
- 70.7–79.4
Keep in mind that mesh ratios can differ from tape-measure values. Our top forehead landmark sits at the mid-forehead rather than at your hairline, which gives a shorter face length than a ruler measurement from the hairline. Because the model fits a standard face to every photo, small score differences, such as 75 versus 79, are within the range of our calibration faces.
Practical ways to frame your proportions in photos and styling
The golden ratio test is a way to explore facial geometry, not a target to chase. The camera setup changes what the photo shows, so it helps to keep it consistent. Photos taken very close to the face can distort relative widths, so hold the phone at arm's length and at eye level for a more even result.
Hair and grooming change how a face looks in a photo, even though the tool measures landmarks rather than styling. Keeping hair off the forehead, brows and jaw lets the model place its points on the actual edges. A neat brow line or a trimmed beard can also make those edges easier to see.
You can test glasses the same way. Take one photo with your usual glasses and one without, in the same light, and compare the results. Frames can cover or shadow the eye and brow points, so the numbers may shift slightly between the two photos.
Reading your golden ratio score
The golden ratio is a popular idea, not a beauty standard: the studies we cite point to other proportions, and the “Phi mask” has been criticised in the research literature. We show the numbers because people ask for them, not because 1.618 is the goal.
| Score | What it means | Keep in mind |
|---|---|---|
| 80 and up | Your ratios sit closer to 1.618 on average | Above the 90th percentile of our calibration faces (79.4) |
| 70–80 | A mix of ratios that are closer to and further from 1.618 | Where most of our calibration faces fall |
| 60–70 | Ratios sit further from 1.618 on average | Below the middle range of our calibration faces; retake with a level head and neutral mouth to check |
| Below 60 | Most ratios are far from 1.618 | Check the photo angle first: a tilted head changes the height ratios |
What the golden ratio test can't tell you
- A higher score is not a more attractive face. The studies cited here do not support 1.618 as the ideal for facial proportions.
- The top point is the mid-forehead on the face outline, not your hairline, so the face length ratio differs from tape-measure charts.
- The face model fits a standard face to your photo, which narrows differences between faces.
- Photos taken very close up can distort relative widths. Hold the phone at arm’s length.
- The calibration range comes from 133 real front-facing photos, not a survey of real human populations.
Photo tips
Face the camera straight on
Hold the phone at eye level and look into the lens. A turned or tilted head changes the distances, and the tool will ask for another photo if the angle is too large.
Even, soft light
Face a window or a soft lamp. Hard side light can shadow one half of the face and make landmark placement less reliable.
Neutral expression
Relax your mouth and eyes. Smiling moves the mouth corners and cheeks, which can change several measurements.
Hair off the face
Brows, forehead edges and jaw need to be visible. Fringes and hair over the cheeks can hide the edges and make the points less accurate.
Questions
References
- Kartynnik, Y., Ablavatski, A., Grishchenko, I., & Grundmann, M. (2019). Real-time facial surface geometry from monocular video on mobile GPUs. arXiv:1907.06724. arxiv.org/abs/1907.06724
- Pallett, P. M., Link, S., & Lee, K. (2010). New “golden” ratios for facial beauty. Vision Research, 50(2), 149–154. doi.org/10.1016/j.visres.2009.11.003
- Holland, E. (2008). Marquardt’s Phi mask: Pitfalls of relying on fashion models and the golden ratio to describe a beautiful face. Aesthetic Plastic Surgery, 32(2), 200–208. doi.org/10.1007/s00266-007-9080-z
- Farkas, L. G., Hreczko, T. A., Kolar, J. C., & Munro, I. R. (1985). Vertical and horizontal proportions of the face in young adult North American Caucasians: Revision of neoclassical canons. Plastic and Reconstructive Surgery, 75(3), 328–337. doi.org/10.1097/00006534-198503000-00005