What facial symmetry means in a photograph
Bilateral facial symmetry means corresponding features on opposite sides of a central line occupy matching positions. In a photograph, that could mean the two mouth corners sit at similar heights and at similar distances from the center. This tool compares selected landmark pairs in that way. It does not compare every pixel, the texture of your skin or the brightness of the two sides of the image.
There are several different questions hidden inside the phrase face symmetry. The shape of the outer contour, the position of an eye, the movement of a smile and the distribution of light are not the same measurement. A face may look more uneven under side lighting even if the selected landmark positions barely change. Conversely, a small movement in a mouth corner can change a landmark measurement without making the overall portrait look noticeably different.
The result describes the alignment of features visible in one image. It is not a comparison against a perfectly mirrored replacement face. Natural faces do not need to behave like identical reflected halves, and a mathematically closer match is not automatically a more appealing expression. Use the individual differences to understand the result instead of interpreting the total as a judgment about your appearance.
How the forehead-to-chin midline sets the comparison
The tool estimates a vertical axis from a forehead landmark to a chin landmark. It then expresses the paired landmarks relative to this axis. This makes the comparison less sensitive to rolling your head toward a shoulder than a simple comparison against the vertical edge of the photograph. The image frame itself is not assumed to be your facial centerline.
For each pair, the calculation combines two discrepancies: how differently the points sit to either side of the axis and how much their vertical positions differ. If the reflected positions match, the difference approaches zero. If one point sits higher or farther from the center than its partner, the difference grows. The combined distance is divided by cheek-to-cheek width and shown as a percentage.
That normalization helps explain the units. A difference of 2 percent means the mismatch equals about two hundredths of the reference face width in the image. It does not mean that 2 percent of your face is asymmetric. A percentage is also not a millimeter measurement. The centerline is an estimated reference, and changing the detected forehead or chin position can influence several paired results together.
The twelve paired regions behind the breakdown
The breakdown checks outer eye corners, inner eye corners, upper eyelids and lower eyelids. It also checks outer brows, inner brows and brow peaks. The remaining pairs cover nose wings, mouth corners, cheeks, jaw corners and chin edges. These twelve comparisons sample different parts of the face rather than relying on one feature or an outline alone.
Read them region by region. If the mouth difference rises in a smiling photo while the eye-corner values remain similar, expression is a plausible contributor to the change. If several eye and brow measurements shift together, inspect whether one side was partially hidden or whether your head turned. Those are possible explanations to investigate, not causes the calculator has diagnosed.
A larger number in one row means a larger reflected landmark mismatch relative to face width for that pair. It does not establish that the region is the most noticeable feature to another person. Different anatomical areas, expressions and viewing conditions can affect perception. The tool gives all twelve pairs equal weight in its summary, so the summary does not attempt to model how people visually prioritize different features.
What the 0–100 symmetry index actually calculates
The displayed index begins at 100 and subtracts four times the average of the twelve percentage differences. The result is rounded to a whole number and is not allowed to fall below zero. For example, an average paired difference of 2 gives an index of 92. An average of 3 gives 88. These examples explain the arithmetic so the number is not a mysterious rating.
This is an application-defined summary, not a clinically validated scale. A result of 92 does not mean that your face is 92 percent symmetrical, that you are in the ninety-second percentile, or that the detector is 92 percent confident. There is no reference population behind the number and no diagnostic cutoff separating normal from abnormal. Comparing scores from different websites may be misleading because they can use entirely different formulas.
Two portraits can receive the same index while showing different regional patterns. One may have small discrepancies spread across many pairs; another may have one larger discrepancy and close agreement elsewhere. The breakdown preserves that distinction. If your interest is in a specific feature, compare its row under consistent photo conditions rather than relying solely on the combined score.
Natural asymmetry and the limits of a normal range
Small visible differences between sides do not by themselves establish a problem. This calculator cannot determine a clinical normal range from a selfie, and it should not invent one from its score. The measurement concerns projected positions in a photograph; it does not assess function, symptoms, history or underlying structures. A numerical discrepancy alone cannot explain whether a difference matters medically.
There is also no universal photograph-based amount of asymmetry that everyone will notice in the same way. A familiar mirror view can feel different from an unmirrored image even though neither shows a new physical change. Mirroring swaps the sides of the picture, while perspective and expression can change the measured geometry. Keep those distinct when looking at a result that feels unfamiliar.
Use the tool for descriptive comparisons and curiosity. Do not infer a disorder, a need for treatment or an attractiveness ranking from a low index or one larger row. If a change concerns you for reasons beyond a photograph, this page cannot evaluate that concern. Its output is limited to the selected landmarks and the formula explained above.
Separating pose and expression from repeatable differences
Begin with a centered camera, relaxed mouth and both sides of the face visible. Keep your gaze forward and avoid turning your nose away from the lens. The centerline adjustment can reduce sideways lean, but it cannot undo a three-dimensional turn. One side of a rotated face is projected differently from the other, which can create apparent asymmetry across several regions.
Take a second neutral photo under the same lighting and distance, then compare the paired rows. If several numbers change substantially, treat the result as sensitive to the photograph. You can also compare a neutral and smiling picture to explore how expression changes the measured positions, provided you describe that as an expression comparison rather than a change in your underlying facial structure.
Keep the original images with any notes you make, since the tool does not save a personal measurement history. Avoid selecting only the highest index from many attempts: that says more about the selection process than about a stable characteristic. A useful conclusion might be that mouth-corner alignment varies with expression while eye-corner alignment is more consistent in your reference photos. That is more specific, and more defensible, than calling a face simply symmetrical or asymmetrical.