Facial symmetry test

A facial symmetry test compares corresponding points on the left and right sides of one photograph. Upload a clear, front-facing photo to see the total, five feature scores and a midline overlay. Read them together. In our camera records, the Mouth score rose slightly while the total fell, so one headline number did not describe every feature's movement.

What a total can hide

On 21 October 2026, we tested four photographs of one consenting participant, with three attempts per photo. The owner recorded the same phone, day and lighting, with approximate distances and angles. Lens, zoom and camera height were not recorded.

The near, front-facing photograph returned 98.6 overall and 98.0 for Mouth on each of its three uploads. The far, front-facing photograph returned 96.1 overall and 98.1 for Mouth on each of its three uploads.

Far minus near was −2.5 points overall and +0.1 for Mouth. Jaw moved from 99.1 to 92.5, a difference of −6.6. These are comparisons between those photographs. This setup changed more than one capture condition, so it cannot isolate a distance effect or establish a physical change.

A refused photo has no score

The other two photos were owner-labelled as turned approximately 15–20 degrees, one near and one far. All three attempts for each photo were rejected. The tool returned:

Your head appears turned. Use a photo facing straight toward the camera.

All six refusals are preserved below. Each has no numerical total or feature result. A refusal gives you a reason to choose another input. It cannot be ranked below an accepted face.

File changes can move features in different directions

On 26 October 2026, we tested five prepared files from a separate public portrait. Each file was uploaded three times in the production Chrome tool. The edits were prepared independently from the original image.

The baseline returned 98.0 overall and 98.2 for Nose. A flat ten-degree image rotation returned 98.1 overall and 97.8 for Nose on all three repeats. The total rose 0.1 while Nose fell 0.4. A JPEG copy returned 97.5 overall and 96.4 for Mouth, compared with the baseline Mouth reading of 97.1.

A higher total therefore did not mean a higher value in every saved field in this test. The rotation was an image-plane edit, separate from the turned-head camera attempts.

Ten additional uploads of the unchanged baseline repeated the displayed total and all five feature values. Each field's displayed maximum minus minimum was 0.0. That describes this file, these runs and the displayed precision. It supplies no ground-truth comparison with anatomy.

Use the result without losing the conditions

Choose one unedited, front-facing photograph with a relaxed expression and even light. Inspect the midline overlay and the five output labels before interpreting the total. Keep the exact file when checking repeatability. If an input is refused, retain the refusal beside the successful runs.

The tool measures estimated photo geometry. It provides no beauty rating or medical assessment. For a separate composite photo feature index, visit the attractiveness test. For the detailed capture setup, see the camera experiment. The file and repeatability experiment explains the prepared-image protocol.

Every saved attempt

These are reused historical observations, recorded on 21 and 26 October 2026. No new measurement was made for this article. “—” means no score was returned. Times below preserve the source records' UTC timestamps.

Original records: camera, prepared files, unchanged-file repeats.