Canthal Tilt Test, One Angle Per Eye
Canthal tilt is the angle between the line joining the two corners of your eye and the horizontal. Upload one front-facing photo and this test reads it for each eye in degrees, tells you which way it leans, and shows the one thing every other calculator leaves out: how far a slightly tilted head moves the number.
Free while in beta, no app needed. The measuring runs on your own device.
Analyze your face
Deterministic geometry, not an AI opinion. Runs entirely in your browser.
🔒 Your photo is read locally and scored on your own device. Nothing is uploaded to score it.
What you get back
One photo in, four numbers out. This is a real run on the example face, not a mock-up: the figures below are what the tool returned for it.
- Average of both eyes
- +5.4°
- Left eye
- +8.1°
- Right eye
- +2.7°
- Left minus right
- +5.4°
- Head roll in this photo
- -2.3°
The average is the number everyone quotes. The other three are what make it readable. Her two eyes differ by 5.4°, which sounds like a lot, but the photo also has 2.3° of head roll, and roll moves the two eyes in opposite directions at about 2° of gap per 1° of roll. That accounts for roughly 4.6° of the 5.4°. The real difference between her eyes is closer to one degree. The section on accuracy below shows the measurement behind that claim.
How your canthal tilt is calculated
You can do this by hand before you trust a tool with it. Two points and a line are all it takes.
1.Take a level, front-facing photo
Eye level, straight at the camera, head level. This step matters more than any later one, and the accuracy section explains exactly how much.
2.Find the two corners of one eye
The inner corner, where the lids meet beside the nose, is the medial canthus. The outer corner, toward the temple, is the lateral canthus. Mark both.
3.Measure the line against the horizontal
Draw a line through the two corners and read its angle against a horizontal line. Outer corner higher than the inner one is a positive tilt; lower is negative. Repeat on the other eye.
The tool does the same thing with less guesswork about where the corners are. It places 478 landmark points on the face and takes the two that sit on the inner and outer corner of each eye. The angle between the line through those two points and the horizontal edge of the photo is the reading. That is the whole calculation; there is no model deciding what your tilt should be.
Two details are worth knowing. The horizontal is the photo's horizontal, not your face's, which is why head roll matters so much. And each eye is measured on its own with the same formula, so the two readings are directly comparable and their difference means something.
Positive, neutral and negative canthal tilt: what each looks like
Three words for three directions of the same line. They describe which corner sits higher, nothing more.
Positive canthal tilt
Neutral canthal tilt
Negative canthal tilt
Positive canthal tilt
The outer corner sits higher than the inner corner, so the eye appears to lift toward the temple. This is the direction most reference ranges for adults fall in, by a few degrees.
Neutral canthal tilt
The two corners sit at roughly the same height and the line runs close to level. This tool calls anything within one degree of level neutral, and says so on the result because that boundary is a convention.
Negative canthal tilt
The outer corner sits lower than the inner corner, so the eye appears to slope downward toward the temple. It is common, it is measurable, and what it means is a separate question this page takes up further down.
The labels are a convention. Where neutral ends and positive begins is not something a face decides; it is a line someone drew. The tool shows the degrees next to the label so you can ignore the label if you prefer.
Typical canthal tilt for men and women
Most published figures agree on the direction and disagree on the exact numbers. A commonly quoted set of reference ranges puts men at roughly 0° to +4° and women at roughly +2° to +8°, with women averaging a couple of degrees more than men. A 2023 study that measured 100 facial points on 42 celebrities reported 8.5° for the women and 6.5° for the men, and cited a general-population figure of about 4°.
The gap between the sexes shows up in every source; the absolute numbers do not. That is expected: each study places the corner points its own way, on its own photographs, with its own way of leveling the head, so the numbers do not transfer between them cleanly. Treat any single range as one measurement convention, not a fact about faces.
What this tool reads on real photographs
Run on 29 official portrait photographs, this tool's average of both eyes had a median of 4.4°, with the middle half of the portraits between 3.3° and 5.3° and the full spread from −0.1° to 10.1°. Twenty-nine faces are not a population and the page does not report a percentile, but they are enough to show where ordinary adult faces land on this particular tool.
How accurate is this canthal tilt number?
Two different questions hide in that one. Is the number right for this photo? Yes: the tool is deterministic, so the same file returns the same degrees every time, and the corner points it uses are the ones you can see marked in the animation. Is the number right for your face? Only as far as the photo is a fair view of it, and for canthal tilt that is a much bigger caveat than for most measurements.
The reason is scale. The entire range of ordinary adult faces spans about ten degrees, and the gap people argue about, between a slightly positive and a slightly negative tilt, is two or three degrees. The measurement is taken against the horizontal edge of the photo. So any tilt of the head in the photo goes straight into the reading at the same scale as the thing being measured. No other calculator we could find quantifies this; the usual advice is a sentence telling you to face the camera. We measured it instead.
What moves the number, measured on 29 real photographs
Head roll: 2° of reading per 1° of roll
Tilting the head sideways raises one eye's outer corner and lowers the other's. Across 29 portraits, each degree of roll moved the left-right gap by 2.0° (±0.1°), almost exactly the geometric prediction. The average of both eyes barely moved (0.03° per degree of roll). Roll is the dominant source of error and it lands entirely in the left-right difference.
Head pitch: about 0.25° per 1° of nod
Dropping the chin lowered the average reading by roughly 0.25° per degree of pitch (±0.1°) across the same 29 portraits. A ten-degree nod, which is easy to do without noticing, is worth about 2.5°, comparable to the entire quoted range for men. This one is a correlation across different faces rather than one face at several pitches, so read it as a magnitude, not a precise constant.
Head yaw: no detectable effect on the average
Turning the head left or right, within the ±11° present in those portraits, produced no measurable shift in the average (0.03° ± 0.13° per degree). That is not proof of zero; it means the effect, if any, is smaller than the difference between one face and the next, which was about 2.5°. Larger turns are rejected by the tool before a reading is given.
Left-right gap: read it with the roll next to it
Among portraits with less than 2° of roll, the two eyes differed by 1.9° on average. Between 2° and 4° of roll the average gap was 6.5°, and above 4° it was 11.4°. A large gap on a rolled photo is the roll; the same gap on a level photo is the face.
What this means in practice. Keep your head level, and check the roll figure the tool prints under your result: it comes from the same face-fitting step and it tells you how much of your left-right gap to discount. Compare only photos taken the same way. And treat the average of both eyes as the sturdy number, because it survives roll; the left-right difference does not.
The tool refuses photographs that are turned, tilted or pitched beyond its limits rather than returning a confident number from a bad input. Inside those limits it still prints the head angles so that you, not the tool, decide how much to trust the reading.
Why your left and right eyes read differently
Almost everyone gets two different numbers, and the first thing to check is the roll figure under the result. A rolled head raises one outer corner and lowers the other, so a 2° roll alone accounts for about a 4° gap between the eyes. Retake the photo with the head level and most of that gap usually disappears.
What remains is real. Faces are not symmetrical, and the eye corners are one of the more stable places that shows up: the outer corners sit at slightly different heights on most people, and the difference is repeatable from photo to photo when the head is held the same way. It is the same asymmetry the face symmetry test on this site measures across the whole face, and a gap of a degree or so at the eyes is ordinary.
Two things do not follow from a gap. It does not mean one measurement is wrong, and it does not mean something needs correcting. Most people cannot see a one-degree difference in another person's eyes at conversational distance; the composite images on the symmetry page, which double every difference, are the only way most people ever notice their own.
Measure the rest of your facial symmetryIs a negative canthal tilt a defect?
If the tool read −2.3° for you, with the two eyes within a degree of each other on a level photo, then the number is right. The same photo will read −2.3° tomorrow, and a different level photo of you will land close to it. Nothing on this page walks that back.
What the number means is a different question, and the research answer is narrower than the version circulating on video. The most-cited study, a 2007 forced-choice experiment (Bashour and Geist, PubMed 17237692), showed judges pairs of female faces where one copy had its inner-corner tilt digitally accentuated, and the more tilted copy was chosen 93% of the time. That is real evidence that the direction of the tilt is something people respond to. It is not evidence for a threshold, for a number of degrees that decides anything, or for a negative tilt being unattractive: every face in that study was compared with an edited version of itself, not with someone else, and nobody in it was rated as a defect.
A 2023 study of 42 celebrities found their tilts a couple of degrees above a general-population figure, which says something about who gets cast and nothing about the ordinary faces that make up the rest of the range. The people writing about this for magazines, including plastic surgeons interviewed on the record, tend to land in the same place: the angle exists and is one feature among many, and a negative one is common in faces widely agreed to be attractive. That is the honest state of things. The angle is real; its power over how you come across is much smaller than the confident version suggests.
If you do want to change it, the next section lists what exists. It is a description of the categories, not a recommendation of any of them.
What people do about canthal tilt
Three categories of approach exist. This page describes what each one changes and stops there: it is not medical advice, it does not recommend a procedure, and it does not link to anyone who performs one.
Makeup and eyeliner
A line that extends upward from the outer corner changes where the eye appears to end, which shifts the apparent tilt without touching the anatomy. It is reversible and it is the only approach here that a photo will pick up as a change in the measured angle only if the liner is mistaken for the corner, which the landmark model generally does not do.
Brow lift
Procedures that raise the outer part of the brow change the tissue above the eye, not the corners themselves. The measured canthal tilt usually moves little; what changes is the impression of lift around the eye. This is a medical procedure with its own risks and recovery.
Canthoplasty and canthopexy
Surgery on the outer corner itself, either repositioning the tendon that anchors it or tightening it. These are the only approaches that change the measured angle directly, and they are surgery, performed on the eyelid, with the risks that implies. Anyone considering them should be talking to a qualified specialist, not to a web page.
Three of the questions people search most in this category are whether the tilt can be changed naturally, whether exercises work, and whether it changes with age. Muscle exercises do not move the tendon that anchors the outer corner. With age the outer corner tends to drop slightly as the supporting tissue loosens, which is why the tilt measured on the same person at 25 and at 60 can differ by a degree or two.
What hunter eyes mean in degrees
Hunter eyes is a bundle, not a measurement. The term, from online looksmaxxing communities, usually means a positive canthal tilt combined with deep-set eyes, a low and heavy brow, and little visible upper eyelid. Canthal tilt is the only part of that bundle a two-point measurement captures, so this tool can tell you the angle and cannot tell you whether you have hunter eyes.
There is no agreed number. Descriptions online typically put the tilt component somewhere in the upper part of the positive range, but the figure varies by who is writing and none of it comes from a study that defined the term. What the measurement can settle is the simpler question underneath: which way your corners lean and by how many degrees. The rest of the bundle is shape and depth, which the related term almond eyes describes from the other direction, by outline rather than by angle, and which this page does not measure.
Questions people actually ask
Now read your own angle
You know what the number measures and what moves it. The reading takes one level, front-facing photo: an angle for each eye, their average, and the head roll to judge them by.
Upload a photoAfter the canthal tilt test
How symmetrical the rest of your face is
The same eye-corner asymmetry measured across five regions, with the mirror halves that make it visible.
The rest of your facial measurements
Eye area, jawline, harmony and symmetry from the same photo, with every input shown.
What face shape the same photo gives you
The seven-shape classification, the ratios behind it, and why the boundaries are conventions rather than facts.