PHOTO GEOMETRY · LOCAL
Face Symmetry Test
See how paired landmarks in this photo sit around a fitted facial midline. The result is a normalized geometry reading, not a judgment or a claim about a fixed trait.
Preparing the analyzer… Your photo will stay on this device.
Fitted midline
The fitted midline follows several central landmarks after correcting in-plane head roll. It is visible in the overlay.
Paired deviations
Six actual left/right pairs are compared: eye corners, brows, cheek edges, mid cheeks, and mouth corners.
What the reading cannot say
A 2-D photo cannot establish skeletal symmetry, attractiveness, health, or how someone appears in person.
HOW THE MEASUREMENT WORKS
A fitted midline, six paired landmarks, and normalized deviation
The calculation
The face symmetry test first corrects in-plane roll and fits a vertical midline through the detected face. It then compares six visible left/right pairs: outer eye corners, inner eye corners, brow points, cheekbones, mid-cheek points, and mouth corners. Each pair is reflected across the fitted axis. The remaining distance is divided by face width, so the result is a percentage of the photographed face rather than a pixel count.
How to read it
A smaller mean mirror deviation means the paired points landed closer after reflection. The largest-pair value identifies the pair most affected in this photo. Neither number is a medical measurement or a verdict on attractiveness: expression, camera angle, hair, lighting, and landmark placement can move the result.
CONTROLLED EXAMPLE
One public fixture, measured repeatedly
In our documented fixture check, the unchanged source photo returned a mean mirror deviation of 0.837%, a largest paired deviation of 1.859%, and estimated roll of −1.945°. A darker copy, a 90% center crop, and a 5° image rotation produced slightly different photo-space readings. That example shows why a face symmetry score belongs to one photograph and why retakes should use comparable conditions.
Read the fixture, transformations, and full observationsIs it facial asymmetry, or the photo?
Head turn
When one side turns away, its visible span changes. A front-photo calculation cannot reconstruct the hidden surface. Retake straight on rather than interpreting the result as a permanent difference.
Head tilt and chin position
The tool corrects in-plane roll using the eyes. It does not fully correct a lifted chin, lowered chin or three-dimensional turn. Keep the camera at eye level.
Camera distance and framing
A close selfie and a more distant portrait can show different apparent proportions. Keep the setup consistent; a center crop changes framing, not the original viewpoint.
Expression, light and occlusion
Smiling moves mouth corners; shadows and hair can affect landmark placement. Use a relaxed expression and even light, and inspect whether the overlay follows the visible features. Passing our quality checks is not a guarantee of a reliable reading.
Read one example across four versions of a photo
| Photo version | Mean deviation | Largest pair |
|---|---|---|
| Original photo | 0.837% | 1.859% |
| Brightness −35% | 0.747% | 1.716% |
| Center crop 90% | 0.933% | 1.878% |
| Image rotation +5° | 0.858% | 1.549% |
A small change in a reading does not identify what changed in the person. This experiment changed pixels in one existing image; it did not test a new camera position or a real head turn. It is not a population study or a calibration dataset. Inspect the source, transformations and full record.
LIMITS AND QUESTIONS
What a face symmetry analyzer can and cannot tell you
Am I symmetrical?
No face is perfectly symmetrical. This tool estimates visible paired-landmark alignment in one front-facing photo; it does not classify a person as symmetrical or asymmetrical.
Why did my symmetry score change?
Head roll, a small turn, expression, camera distance, light, blur, and occlusion can all move landmarks. Compare only photos taken under similar conditions.
Does the test measure my side profile?
No. Projection, profile angles, and other depth-dependent traits are not scored from a front photo. Unsupported traits are left unmeasured rather than guessed.
Is facial symmetry the same as attractiveness?
No. Symmetry is one measurable geometric property. A photograph cannot capture personal preference, movement, expression, voice, or the rest of how people perceive a person.
For scoring weights and reference bands, read the methodology. To see how measurements become styling decisions, open the sample report.