Golden Ratio Face Calculator

Proportions here come from the canon that phi-based analysis has traditionally used. Each is reported as a raw value alongside its distance from the target.

• Face length over face width divides hairline-to-chin by bizygomatic width.
• Mouth width over nose width divides commissure span by alar span.
• Interocular over alar width divides the inner eye corner gap by that same span, and targets unity rather than phi.
• Lips to eyebrows over nasal length divides the lip-line-to-brow distance by nose height.
• Vertical thirds split hairline to brow, brow to subnasale, and subnasale to chin, which the classical canon expects to run approximately equal.
• Horizontal fifths cut the face into five vertical segments about one eye wide, running ear, eye, intercanthal gap, eye, ear.
• Lower face division splits subnasale to chin at the stomion, with the upper part relating to the lower by phi.

Each of those feeds a single arithmetic step.

Scores start from proportional error. Every measured value is compared against its target, the absolute difference is divided by that target, and the result is an error independent of units.

Errors then get averaged with weights attached. Ratios built on the eye corners and alar rims count for more than ratios depending on the hairline, because a hairline is a soft boundary that recedes, hides under hair and cannot be located to the same precision. The weighted mean maps onto a 0 to 100 percentage where zero error gives 100.

The right-hand column is the important one. Real faces scatter widely around every target, and sitting outside the span on two or three rows is unremarkable. Which is exactly why the rows deserve more attention than the headline figure.

Rows tell you which of your proportions sit near the classical reference and which sit far from it. That's genuinely useful if you're weighing hair length, brow position, or where to place emphasis with makeup.

The aggregate hides all of it. Two faces can score identically on completely different underlying patterns, because averaging is designed to throw away distribution. Read down the column instead.

Note also which rows respond to styling. The vertical thirds are bounded at the top by the hairline, so a fringe, a receding hairline or a high ponytail moves that measurement substantially while the bone underneath stays put. Photography moves the others.

Capture requirements match those for symmetry measurement, for the same reason. This is a proportional reading taken from a projected image, so whatever distorts the projection corrupts the arithmetic.

Face the camera squarely at eye level, expression neutral, mouth closed. Pull all hair back until the hairline is visible, since several proportions depend on it and an estimated hairline is the largest single error source in the whole calculation. Light the face evenly from the front with no strong shadow crossing either side.

Stand at least an arm's length back and crop in rather than stepping closer. A front-facing phone lens is short focal length, and it enlarges central features relative to the outline. That inflates nose width and shrinks apparent face width, which corrupts three proportions at once. Even a clean capture leaves the deeper limits in place.

Perfect proportion is not normal, and it is not a target. No real face matches phi on every row. A score of 100 would describe a face that does not exist, and faces adjusted toward it in software tend to look synthetic rather than beautiful. If you take one thing from this page, take that.

The reading is two-dimensional. Projection depth, how far a chin and cheekbones come forward, is absent from a frontal photograph, and profile proportion carries a large share of how a face reads in person.

The targets are conventions. They came out of one aesthetic tradition and were validated mainly against faces from a narrow set of populations. Applying them universally imports an assumption that a single proportional standard suits everyone, which the evidence does not support.

Finally, this is curiosity rather than assessment. A number here says nothing about health, character or worth, and it is never a reason to change your face. The interesting part is the arithmetic, and seeing your own proportions written out.

A drawing convention, measured precisely. Phi at 1.618 supplies a fixed reference, the seven proportions run from face length against face width down to the lower face division at the stomion, and the score comes from weighted proportional error mapped onto a 0 to 100 curve. The rows beat the total because averaging throws distribution away, the hairline stays the weakest landmark in the set, and capture demands a level square frame at arm's length. Perfect proportion remains neither normal nor a target. Open the golden ratio face calculator and read your proportions row by row.