04 / A proportion, not a beauty law
Golden Ratio Face Test
A golden ratio face test should show what it measures. Here you can compare a visible face length-to-width ratio with the product-selected 1.618 reference, inspect the raw value, and understand how the match score is calculated. The reference is optional geometry, not a scientifically established ideal for every face. Your photo stays on this device.
- Photos stay in your browser
- Free to use
- No account required
A geometric reference match, shaped by product-selected rules. It is not an objective measure of beauty or your worth.
Start with one front-facing photo
Drop your photo here or choose a file
JPG, PNG or WebP · up to 30 MB · one adult face
Photo bytes and facial landmarks stay in this browser tab. We do not upload them, store them in localStorage, or attach them to model downloads.
The detector loads asynchronously after you start. The first run may take longer; your photo is not sent with the model request.
The measured ratio beside its reference.
—/100
Choose a photo, then start your local analysis.
This view isolates the visible proportion dimension. It compares mesh-top-to-chin length with cheek width, using 1.618 as a selected reference.The 1.618 formula explained
The golden ratio is commonly written as approximately 1.618. This tool does not search for it across every feature. It measures the pixel distance from mesh point 10 to chin point 152, divides that by cheek width between points 234 and 454, and compares the result with 1.618. The top mesh point is not the hairline, so this measurement must not be presented as a full anatomical face-height assessment.
How the facial proportion check becomes a score
The proportion score starts with relative deviation: the absolute difference between your measured ratio and 1.618, divided by 1.618. A published piecewise curve converts that deviation into a zero-to-one-hundred reference-match value. Smaller deviation gives a higher match. The overall bands below describe the broader four-dimension method; the isolated proportion result is not the same as the complete score.
- 0–<60Larger deviation from the selected ratio reference. Check the endpoints and photo conditions.
- 60–<75A mixed match to the selected ratio reference. Read the raw ratio before comparing photos.
- 75–<87A broad match to the selected ratio reference under this product’s mapping.
- 87–100A close match to the selected ratio reference. This is not a beauty or accuracy percentage.
Read the reference without chasing it
A face can differ from 1.618 and still be expressive, distinctive and attractive to other people. Our selected target is a way to make one calculation inspectable, not evidence that everyone should look alike. Camera perspective and head position also change the ratio. Review the raw value, use a consistent photo if comparing, and explore the broader measurements rather than optimizing your appearance around one proportion.
Questions, answered
What is a golden ratio face test?
It is a comparison between chosen facial measurements and a numerical reference associated with the golden ratio. In this tool, the selected measurement is visible mesh length divided by cheek width, compared with 1.618. Other products may measure different distances. Check the exact formula before comparing their results, because the same test name does not guarantee equivalent definitions.
How does the golden ratio face calculator turn a ratio into a score?
It compares the visible mesh length-to-cheek-width ratio with the selected reference of 1.618, calculates relative deviation, and maps that deviation through the published product curve. A ten-percent deviation maps to 85 under this rule. These are reference-match numbers, not beauty percentages. Try the calculator above, then open the formula details to inspect your photograph’s inputs.
Which points are used for the facial ratio?
The length uses landmarks ten and one hundred fifty-two, while the width uses landmarks two hundred thirty-four and four hundred fifty-four. Coordinates are converted to pixels before distances are compared, avoiding a non-square-image ratio error. The top landmark belongs to the detected mesh rather than the actual hairline, and the width is a visible cheek proxy.
How is the 1.618 match calculated?
We calculate the absolute difference between the visible ratio and 1.618, then divide by 1.618 to obtain relative deviation. The published curve assigns reference-match values at defined deviation points and interpolates between them. You can inspect the formula in the result. The curve is a product choice, not a population-calibrated probability of being attractive.
Why is my golden ratio result different elsewhere?
A different site may use a hairline estimate, face bounding box, another landmark model or a different scoring curve. Even the same image can then produce a different measurement. Compare distance definitions and reference targets before comparing scores. A number with the same label is not automatically measuring the same thing, and neither site becomes ground truth by agreeing.
Can a selfie lens change the proportion reading?
Yes. Perspective depends on camera distance, and a close lens can make nearer facial features appear larger. Turning or tilting the head also changes the visible proportions and landmark placement. Use a front-facing photo taken from a comfortable distance with even light. The test cannot reconstruct a distortion-free three-dimensional face from a single ordinary photograph.
Are facial thirds and fifths included here?
They are available as separate dimensions in the full breakdown, while this page highlights the chosen 1.618 proportion reference. The vertical segments use visible mesh boundaries, not a measured hairline. Horizontal fifths use projected landmark spans. Keeping the dimensions separate makes their definitions easier to inspect and avoids claiming that every facial relationship should equal the golden ratio.
Can I use the result without sharing a photo?
Yes. This free check runs inside your browser and requires no account. The picture and facial landmarks remain in tab memory rather than being uploaded or saved in localStorage. If you choose to export the result, the JSON contains measured readings and method information, without photo bytes, the original filename or the detected landmark coordinates.