Facial Symmetry Test
Upload one photo. See your symmetry score and both mirrored halves of your face.
Your photo never leaves your device.
How symmetrical is my face? Reading your score
Your score is not a grade, and a number in the thirties is not a failure. It is a position in a distribution, and most people sit lower on it than they expect.
Under 0.6% deviation. Rare. About 1% of scans land here.
Roughly 0.6% to 1.7% deviation. About 21% of scans land here.
Roughly 1.7% to 3.4% deviation. Most people are here. About 42% of scans land here.
Roughly 3.4% to 5.3% deviation. About 22% of scans land here.
Over 5.3% deviation. Usually the photo. About 14% of scans land here.
The median score on this tool is around 38. Some symmetry tests report almost everyone in the 70s and 80s, which sounds kinder but compresses every face into the same narrow band. We would rather give you a number that actually separates you from the person next to you, and then tell you honestly where it sits.
What facial symmetry actually measures
Symmetry is the difference between the two halves of your face, measured against a line running through the bridge of your nose and the centre of your chin. If your left eye corner sits 3mm from that line and your right sits 4mm, the difference is 1mm. We do that for four mirrored pairs — outer eye corners, mouth corners, cheekbones and brows — average the differences, and express the result as a percentage of your face width.
That percentage is your deviation, and it is the real measurement. The score out of 100 places that deviation on a curve so small differences at the symmetric end still separate clearly.
Both numbers are shown, because the deviation is what was measured and the score is what tells you where you stand.
Reading your mirrored halves
The two composites take one half of your face and reflect it onto itself. The left composite is what you would look like if both sides matched your left side; the right does the same with your right.
They will look strange, and they will look different from each other even at a high score. That is arithmetic rather than evidence: mirroring doubles every difference, so 2mm becomes 4mm. It doubles the lighting too, which is why a face lit from one side produces one bright composite and one dark one while the underlying geometry is nearly identical. Hair parted to one side gets duplicated as well.
Read the images for intuition and the number for the measurement. When they seem to disagree, the number is the one that was measured.
Why your score changes between photos
Most low symmetry scores are photographs, not faces. Turn your head twenty degrees and the camera compresses the far side of your face; the measurement reads that compression as asymmetry. In our own testing, one face measured 0.4% deviation square-on and 28.5% in a photo taken moments later at an angle. Same face, seventy times the reading.
Head tilt is the one thing you do not need to worry about here. We measure the angle between your pupils and rotate the landmarks level before measuring, so a photo tilted eight degrees scores within a tenth of a point of the level version. Most symmetry tools skip this step, which is part of why the same person gets wildly different results across different sites.
Four things move your score, in this order:
- Face the camera square-on. By far the biggest source of error. If your head is turned, we say so above your result.
- Stand back about a metre. Close selfies magnify whichever half of your face is nearer the lens.
- Face the light, do not stand beside it. Side lighting barely moves the number but makes the composites hard to read.
- Keep a neutral expression. A natural smile is almost never even, and it moves the mouth corner landmarks.
Symmetry and attractiveness
Symmetry does correlate with perceived attractiveness in the research literature, but the effect is smaller than the internet suggests and it is only one input among many. Faces widely considered striking are frequently asymmetric, and digitally perfected symmetry tends to read as artificial rather than beautiful.
Essentially every human face is asymmetric to some degree. The two halves develop semi-independently, and habit reinforces the difference — chewing on one side, sleeping on one side, a dominant eye. A measurable difference between your halves is the normal condition, not a flaw to be corrected.
Frequently asked questions
Is my face symmetrical?
Almost certainly not, and neither is anyone else's. The two halves of a face develop semi-independently, so a measurable difference is the normal condition. The useful question is not whether you are asymmetric but by how much: most people measure between roughly 1.7% and 3.4% deviation, which scores 30 to 54 on this tool.
What is a good facial symmetry score?
Anything above 55 puts you in the more symmetric fifth of scans. Above 80 is genuinely rare, around 1%. The middle band, 30 to 54, holds about 42% of everyone who uses this tool. Bear in mind that other symmetry tests use gentler curves, so a 78% elsewhere and a 40 here can describe exactly the same face.
Why do my two mirrored halves look so different?
Mirroring doubles every difference, so a 2mm gap becomes 4mm. It doubles the lighting as well: if one side of your face was closer to a window, that side is brighter in both halves of its composite. Hair parted to one side gets duplicated too. The composites exaggerate by design. The number is the measurement; the images are the illustration.
Why did I get a different score on another symmetry test?
Three reasons. Different tools use different landmark sets, so they are not measuring the same points. They map their measurements onto different curves, and most use a gentler one than ours. And most do not correct for head tilt, so a slightly tilted photo scores badly on those tools and normally on this one.
Does a turned head affect my symmetry score?
Heavily. A head turned twenty degrees compresses the far side of your face and the measurement reads that as asymmetry. We check for this and tell you when it has happened. Head tilt is different: we rotate the landmarks level before measuring, so tilt costs you nothing.
Is a perfectly symmetrical face more attractive?
Less than people assume. Symmetry correlates with perceived attractiveness in research, but weakly, and it is one factor among many. Perfectly symmetrical faces tend to look artificial rather than striking, and plenty of faces widely considered beautiful are visibly uneven.
Can I improve my facial symmetry?
Bone structure is fixed without surgery, and that is most of your score. Soft-tissue asymmetry from posture, jaw clenching or sleeping position can shift slightly over months. Be sceptical of facial exercises marketed as symmetry fixes: in practice, the largest measurable change most people see comes from taking a better photo.
Why is nobody perfectly symmetrical?
Faces grow as two halves that develop semi-independently, and daily habits reinforce the difference: chewing on one side, sleeping on one side, one dominant eye, one dominant hand. Perfect symmetry would be the anomaly.
How is the score calculated?
We detect 478 facial landmarks in your browser, draw a midline through the bridge of your nose and your chin, then compare four mirrored pairs against it: outer eye corners, mouth corners, cheekbones and brows. The average difference, as a percentage of your face width, is your deviation. That is placed on a decay curve to produce the score out of 100.
How many times can I test?
As often as you like. There is no account, no limit and no charge, because the whole measurement runs on your own device and costs us nothing. Trying two or three photos is worth doing: the spread between them tells you how much of your score is your face and how much is your camera.
Do you store my photo?
No. Landmark detection and the mirrored composites are all generated inside your browser using canvas. Nothing is uploaded, nothing is transmitted and nothing is saved. Close the tab and it is gone.