Measurement reference · FaceFluent editorial · September 13, 2026
Mirrored-landmark deviation
How far paired landmarks sit from their mirror positions about an estimated midline, averaged and expressed as a percentage of interpupillary distance. Zero would be a perfectly mirrored photograph.
Educational method only. Personal numerical values are currently unavailable in the free preview and reports under the current offer. Existing purchases retain their purchased scope.
Also called mirrored-landmark deviation, facial asymmetry.
Reference range
None.
We have not validated a population reference interval for this photographic method. Published anatomical measurements do not automatically validate mesh-based estimates.
How we calculate it
For each mirrored landmark pair we compare the distance of each point from the midline and the difference in their heights, take the average across all pairs, and scale it by interpupillary distance.
What moves the number
Almost everything. A degree of head turn, a lopsided smile, uneven lighting on one cheek, or a single mislocated landmark will all move it. We have not established a repeatability interval for this engine, so small differences cannot be interpreted as anatomical change.
What it cannot tell you
We publish no reference band and no score. Perfect symmetry is not a property of human faces, small asymmetry is universal, and a number derived from one photograph cannot distinguish a real asymmetry from the way you were standing.
Common questions
Is anyone’s face perfectly symmetrical?
No. Small left-right differences are universal, and a face that mirrored exactly would be unusual rather than ideal. The number this method produces is a description of one photograph, not a distance from a target.
Does a facial symmetry test measure attractiveness?
No, and we do not publish one as if it did. This method compares where mirrored landmarks sit and reports nothing about how a face is judged. Tools that turn the same geometry into a percentage score are adding a judgement the geometry does not contain.
Why did my symmetry result change between two photos?
Almost anything moves it: a degree of head turn, a lopsided smile, uneven light on one cheek, or one mislocated landmark. We have not established a repeatability interval for this engine, so a difference between two photographs cannot be read as a change in your face.
References
None. We have not found a published population range for this measurement, so there is nothing here to cite. We would rather show an empty list than a decorative one.
FaceFluent describes mirrored-landmark alignment in a photo without turning it into an attractiveness score.
Read: Symmetry without a scoreTry a haircut on your photo. On-device landmark annotations are optional and stay in this browser.
Try a haircut on my photo Try free face analysis on your deviceThis page is educational. It is not a clinical assessment, a diagnosis, or a claim about how you look. Read the methods and limitations.