AI Canthal Tilt Test: Positive, Negative, or Neutral?

This AI facial analysis measures that angle in degrees from your photo, per eye, on your device. An angle is a description, not a verdict.

The Short Version

Canthal tilt is a small number measured between two landmarks per eye: the inner corner and the outer corner against level. Degrees, not vibes. A positive value means the outer corner sits higher, a negative one means it sits lower, and the scan reports the angle it actually measured on your photo.

What Is Canthal Tilt?

The medial canthus is where your upper and lower lids meet beside the nose. The lateral canthus is where they meet at the outer edge, anchored to the orbital rim by the lateral canthal tendon. The line joining them is the intercanthal axis, and its angle against the face horizontal is the measurement. That reference runs across your face, not across the photograph, which is why a tilted camera changes the picture and not the eye. Eye tilt, canthal angle and palpebral fissure inclination name the same quantity.

What Is a Positive Canthal Tilt?

A positive canthal tilt means the outer corner of the eye sits higher than the inner corner, reported here as a positive number of degrees. The sign covers everything above zero, so 1 degree and 9 degrees are both positive tilt and are not the same eye. Most published adult averages sit above zero. Upturned eyes is the mainstream word for it.

What Is a Negative Canthal Tilt?

A negative canthal tilt means the outer corner of the eye sits lower than the inner corner, reported here as a negative number of degrees. The sign describes direction, not severity: negative tilt starts anywhere below zero and runs from a fraction of a degree to a clearly downward set. Downturned eyes is the everyday word for it. A negative reading records where two landmarks sat in one photograph, and the studies that weigh that come later.

What Is a Neutral Canthal Tilt?

A neutral canthal tilt means the two corners sit level, or close enough to level that the measurement cannot separate them. Near is doing real work there: a 2D estimate close to zero crosses the boundary on camera height alone, so a neutral result and a slightly positive one are often the same eye twice.

Positive vs. Negative Canthal Tilt: What's the Difference?

The table sets the three types side by side, with the sign each carries and what each reading leaves unsaid.

Positive and negative canthal tilt differ in one thing: the sign of a single angle. The gap between the two categories is often smaller than the error in one photograph, which makes positive vs negative canthal tilt a finer comparison than the labels suggest. Someone at plus 1 degree and someone at minus 1 degree share more eye geometry than two people who both measure plus 6. Shot from below, positive canthal tilt vs negative swaps inside one sitting, and negative canthal tilt vs positive swaps back at eye level.

How Your Canthal Tilt Is Measured, in Degrees

This test measures canthal tilt from four eye-corner landmarks and five midline points on a 478-point face mesh. The mesh runs inside your browser, which is why photos never leave your browser. A face-horizontal axis is fitted through the midline points, and each eye's intercanthal line is projected against it. The result comes back per eye and averaged, beside the raw uncorrected angle, so the roll correction is visible rather than asserted.

You can measure your own canthal tilt without any tool, using a phone, a straight edge and four steps.

1. Photograph your face straight on with the lens at eye height, chin neutral, hair clear of both outer corners.
2. Mark the inner corner and the outer corner of one eye, where the upper and lower lids meet.
3. Draw your reference horizontal along the line between your two inner corners, not along the edge of the photo.
4. Measure the angle between the corner-to-corner line and that horizontal, then repeat on the second eye and average the two.

Both routes return a 2D photo estimate against a fitted face axis. Neither is the clinical measurement, which is read against the Frankfort horizontal plane with the head positioned, and a photo estimate does not replace it.

What Is a Normal Canthal Tilt? Average Angles by Sex and Population

Two conventions are in play here, and mixing them is the most common error online. Fry reports tilt directly, in degrees of inclination above a horizontal drawn from the inner corner. Gao reports a three-point angle where 180 degrees means level, so 177.6 degrees is 2.4 degrees of tilt. Both point the same way: East Asian samples measure more upward tilt than white samples, and female means sit above male means.

Every row names its population and its sample size, because a degree value without its sample is the error this page exists to avoid. Liu sets those celebrity means against a 4.12 degree Caucasian average reported by Rhee.

Canthal tilt in men and women

The measured sex difference is real and small. Liu and colleagues measured 21 female and 21 male Caucasian celebrities and reported 8.50 degrees of canthal tilt in the women against 6.51 degrees in the men, at P below .010. Those 42 faces were picked for their attractiveness from editorial rankings, which makes them selected faces rather than a norm. Gao's population sample points the same way with a smaller gap: 2.4 degrees for Caucasian women, 0.8 for the men.

What the celebrity datasets actually measured

The celebrity numbers circulating online come from one paper. Liu, Salinas and Sharaf ran 100 facial landmarks across 42 Caucasian celebrities aged 23 to 42, selected from editorial rankings between 2017 and 2022, from frontal photographs with no facial animation. It licenses one claim: that this group of selected faces measured these angles. A correction in November 2023 fixed an error in male interpupillary distance and left the tilt values standing. We do not publish the canthal tilt of any named person.

What the Evidence Says About Canthal Tilt and Attractiveness

Bashour and Geist produced the strongest single effect in this literature: female faces with an accentuated medial canthal tilt were preferred 93% of the time over the unmodified original, at p below 0.0001. Judges made a forced choice between two versions of the same face, one edited in Photoshop. That licenses one claim, that raters detect and prefer the edit, not the claim that your tilt sets how attractive your face is.

Kempa and colleagues tested the opposite direction and found the opposite sign. Across 2,081 participants rating nine edited versions of one unisex face, raising the lateral canthus lowered ratings of attractiveness, femininity and trustworthiness, and raised dominance, threat and disgust. Two careful studies pointing opposite ways is the honest state of the question.

Sforza and colleagues compared groups instead of edits, digitizing landmarks on 126 attractive Italian subjects recruited at beauty competitions and 199 normal subjects aged 18 to 30. The attractive group measured increased eye fissure inclination, reduced intercanthal width and longer eye fissures. The paper reports z-score patterns rather than degrees, so it supports a group difference, not a threshold a face passes.

Vaca and colleagues bound all three findings with a camera. Twelve volunteers were photographed with the Frankfort plane set between minus 15 and plus 15 degrees, and 11 blinded evaluators ranked the results. Downward head tilt tracked with better aesthetic ranking at rho = 0.901, and positive intercanthal tilt grew more pronounced as the head tilted down, at P below 0.001. Head position moves the angle and the rating together.

Is a Negative Canthal Tilt Bad? Is a Positive One Better?

A negative canthal tilt is ordinary anatomical variation, and neither direction is better in any sense a measurement supports.

The population data puts the question in proportion. Published adult means sit within a few degrees of level, with standard deviations of roughly 2 to 3 degrees. Most of any measured group lands close to zero, and the rest spreads to both sides. The question became common through a short video format that pairs a filter with a verdict, and the filter has no reference population. The research that exists reports preferences in rating tasks on edited images: a face is altered, raters rank the versions, the difference is recorded. That is a finding about edits and raters, not a rule about faces, and it says nothing about how a face reads across a conversation.

Hunter Eyes, Upturned, Downturned: The Other Words for Tilt

Upturned and downturned are the mainstream words for positive and negative canthal tilt, and they map onto the measurement cleanly. Hunter eyes does not: it bundles four traits, and canthal tilt is one of them. The other three are low eyelid exposure, deep-set orbits and a close brow, all named here as not measured. A separate guide covers looksmaxxing, and what the terms mean.

• Upturned eyes: the mainstream word for a positive canthal tilt, same measurement, different register.
• Downturned eyes: the mainstream word for a negative canthal tilt, and the phrasing most makeup guides use.
• Hunter eyes: a bundle of four traits rather than a synonym, of which tilt is the measured one.

Can You Change Your Canthal Tilt?

You cannot change your canthal tilt at home, because the lateral canthal tendon anchors to the orbital bone and neither one responds to exercise. Three tiers of claims circulate, ranked here by what the evidence supports.

• What changes how the tilt reads: eyeliner for a negative or positive canthal tilt, lash placement and brow shape, which work immediately and change the read rather than the angle.
• What changes nothing: eye exercises, squint training, facial massage and mewing claims applied to the eye region, none with published evidence of moving a canthal tendon.
• What sits outside this page: canthoplasty and canthopexy, real oculoplastic operations that belong with a qualified surgeon after an examination, with no procedure or cost detail here.

The honest answer to fixing a negative canthal tilt without surgery is the first tier and nothing past it: the read changes, the angle stays.

What This Score Cannot Tell You

The band derivation and the failure modes sit in how we measure, and where the measurement fails.

You Know Your Type: What Should You Do With It?

The type you just found out you are describes one photograph, and it raises three practical questions: whether the reading holds, what can change, and what cannot. Confirm the number before you act on the label. The three sections below take those in order.

Your band names which of the three types the angle falls into. Degrees are not the site's 1 to 10 scale, which what a 1-10 attractiveness score really means covers separately.

How to confirm it

1. Retake the photo with your rear camera, lens at eye height, chin neutral, in the same light as the first.
2. Keep your hair off both outer corners and your expression flat, since squinting moves the outer corner landmark.
3. Measure the new photo and note both averages along with the roll delta the result shows.
4. Compare the two angles rather than the two labels, and read the spread between them as your real margin.

Three photos give a more honest spread than two. When they disagree, the photograph is the variable.

What it changes

Makeup and brow placement change how the tilt reads in a photograph and to a viewer, and nothing else. Liner drawn past the outer corner and angled upward extends the visual line of the eye. Liner that thickens toward the outer third does it more quietly. Brow shape sets the frame the corner is read against. All three are changes to the read, not to the angle: measure again with the makeup on and the number is unchanged. A separate guide covers the eyeliner techniques, step by step.

What it does not change

The angle itself stays put. The lateral canthal tendon attaches to a bony tubercle inside the orbital rim, and bone and tendon answer to neither exercise, massage nor routine. Soft tissue around the lateral canthus does shift slowly across decades. Canthoplasty and canthopexy reposition the tendon surgically, the only entry here that moves the measurement, and they belong with a qualified surgeon.

Is the Canthal Tilt Test Accurate?

The canthal tilt test is reproducible: the same photo and the same version return the same angle, which is a narrower claim than being right about your face.

Three design choices carry that claim. The result prints the raw image-horizontal angle beside the roll-corrected one, so you watch the correction happen instead of trusting it. The instrument refuses an angle above 5 degrees of pitch or 8 degrees of yaw. It refuses again when hair, lashes or a glasses rim crosses a corner, or when the face sits too small in frame. And what it reports is a 2D photo estimate against a fitted face axis, not a clinical measurement against the Frankfort plane. A percentile and a confidence band stay off the screen until a named reference study exists. The rest is in the method, the version and the test-retest data.

Eye Region Metrics: What Else Is Worth Measuring

The eye corner is one measurement in a region with several: brow position, eye spacing, eyelid exposure and orbital depth. None of them is the canthal angle.

The eye shape detector answers what your eye shape is by classifying the outline of the lid opening: almond, round, hooded or monolid. This page measures the corner angle instead.

The full AI face analysis scores every component at once, and the full attractiveness test is the namesake instrument: 478 landmarks, five components, one 1-10 result.

Not medical advice. This is a measurement of a photo, not a diagnosis, a clinical assessment, or a verdict on you.