Canthal Tilt: What It Is and How to Measure Yours Properly

Each eye has two corners. The inner corner (the medial canthus) sits next to the nose, and the outer corner (the lateral canthus) sits toward the temple. Draw a straight line between them and compare it with a horizontal line. That angle is the canthal tilt. Clinicians usually call it the palpebral fissure slant or palpebral fissure inclination, which is the term you will see in the studies below.

The sign is a convention. When the outer corner sits higher than the inner corner, the angle is positive and the eye reads as upturned. When the two corners are level, it is neutral. When the outer corner sits lower, it is negative and the eye reads as downturned. Your two eyes are measured separately, and they are rarely identical.

It is not the same thing as the shape of your eyelid, the fold above it, or how open the eye is. Those change how the eye looks, but the measurement only uses the two corner points. That is also why an expression matters so much: smiling or squinting moves the lower lid and the outer corner, so the angle changes even though nothing about your anatomy has.

Most canthal tilt calculators ask you to click two corners on a photo and then measure the angle against the edge of the image. That works only if your head was perfectly level when the photo was taken. If your head is rolled to one side by 3°, one eye gains 3° and the other loses 3°. Averaging the two eyes cancels the roll, but the per-eye numbers, and any left-versus-right comparison, are simply wrong.

FaceCheckup handles this differently. It finds 478 landmarks on your face with Google's MediaPipe Face Landmarker, then rotates them so the line between your pupils is horizontal before it measures anything. It also uses the 3D landmarks to rotate a slightly turned or tipped head back to a frontal view. Only then does it measure the angle from each inner corner to each outer corner, so the reference is your own face, not the edge of the photo.

Use this routine for a reading you can trust:

Head roll is the obvious error, but chin position matters too. The outer eye corner sits further back on the head than the inner corner, so tipping the chin down raises the outer corner in the picture and tipping it up lowers it. Camera distance adds a smaller shift of its own.

To show the size of these effects, we took Google's MediaPipe canonical face model (the 3D template face the landmarker is built around), rotated it, projected it as a camera would see it, and measured the tilt the same way a basic calculator does. These are geometry results for one template face, not measurements of real people, and the template's own angle is not a population norm. What matters is the change from the straight-on row.

Researchers have measured this angle in large groups, and the clearest finding is that there is no single normal value. Averages differ by population, by sex and by method. Here are the studies we read, with their exact figures.

Choi and Park (2002) measured 996 Korean adults and reported 8.5° ± 2.0° for men and 8.8° ± 2.5° for women, with no significant difference between the sexes. The abstract does not state which reference line was used. Packiriswamy, Kumar and Rao (2012) photographed 300 Malaysian South Indian adults aged 18 to 26 and measured the angle of the line through the inner and outer corners against a horizontal reference, which is the same definition FaceCheckup uses. Men averaged 4.71° ± 2.39° and women 7.05° ± 3°. Vasanthakumar, Kumar and Rao (2013) found 5.053° for men and 6.102° for women in 200 South Indian adults.

No. In every study above, the group average is positive, from 1.2° in one quoted Indian sample of women to 9.64° in another quoted Indian sample. So a positive canthal tilt is the typical finding, not an unusual one. What none of these papers reports is the share of people with a positive, neutral or negative tilt, so any percentage you see quoted for that has no source we could find.

A rough way to read your own number: within one standard deviation of the averages in the two large studies we read in full, values run from about 2° to about 11°. FaceCheckup shows that span as context next to your result (group averages 4.7° to 8.8°, about 2° to 11° within one SD). It does not score you on it, because the right comparison depends on which population you belong to and these are only a handful of groups.

Forum lore often names an 'ideal' range, such as 5.5° to 8.5°. We could not find any peer-reviewed study that sets an attractiveness-optimal tilt in degrees, for women or for men. Treat those ranges as opinions.

A neutral tilt means the two corners sit at roughly the same height. A negative tilt means the outer corner sits lower. For giving style tips, FaceCheckup sorts readings above +2° as positive, below −2° as negative and anything in between as neutral. That cut-off is our own setting, chosen because readings closer to zero than that are within measurement error. It is not a medical threshold.

If your scan says negative, read our guide to negative canthal tilt before you worry. It covers how often the photo is the cause, what the averages say, and the makeup and brow techniques artists use for a downturned outer corner. It also gives the honest answer on 'fixing' it.

There is one widely cited experiment. Bashour and Geist (2007) showed people pairs of female faces, one unmodified and one edited in Photoshop to accentuate what the authors call medial canthal tilt. The edited faces were preferred 93% of the time (p < 0.0001). That is a strong result for one manipulation, but it was a forced choice between two versions of the same face. It used female faces only, and the abstract does not give a degree value that worked best.

Two broader findings keep this in proportion. Hönekopp (2006) found that individual, private taste explains about as much of attractiveness judgments as shared taste does. And camera distance alone shifts how the same face is rated (Bryan, Perona and Adolphs 2012). A single angle on a single photo cannot tell you how other people see you, which is why FaceCheckup reports tilt as a measurement and never as a rating.

Makeup artists don't try to change a tilt. They place liner and brow weight to follow it or to balance it, depending on the look you want. Every tip below is a makeup-artist convention drawn from at least two published sources, and the sources sometimes disagree, which we note.

If your outer corners lift (positive tilt), artists at Makeup.com suggest tightlining the upper lash line and following the eye's own upward curve with a small wing, keeping liner thin at the inner corners. Clinique takes the opposite approach and adds a slightly downward tail to balance an upturned eye. Both are valid choices.

If your corners are level (neutral tilt), liner is a style choice. A classic wing that follows the lower lash line outward works, and you can pick the length. Only one source covers neutral tilt directly, so treat this one lightly.

If your outer corners dip (negative tilt), the most consistent advice from Charlotte Tilbury and the Makeup.com artists is to keep the upper liner thin at the inner corner, thicken it over the outer third and flick it slightly up and out. Start the wing higher than the natural lash line, and judge it with your eyes open. Keep heavy dark liner off the outer lower lash line, or smudge it upward. For brows, L'Oréal Paris and HighBrow Beauty both suggest brushing and filling the tails upward so the tail never sits lower than the brow head.

We looked for optician guidance that matches frame shapes to canthal tilt and didn't find any with two independent sources, so we don't give one. Frame advice on FaceCheckup comes from your face shape instead. For close-set eyes, opticians suggest a clear, light or thin bridge, though that convention is only weakly sourced. The face shapes guide covers frames for each outline.

The same scan also measures eye spacing, eye openness and brow height. Brows frame the outer corner more than anything else, so if you want to change how your eyes read, the eyebrow shapes guide is the most useful next step. For overall left-right balance, the facial symmetry test shows whether a difference between your two eyes is larger than the normal small mismatch every face has.

The result shows your canthal tilt averaged across both eyes, plus a separate 'tilt difference, left vs right', and the Eye tilt layer draws each eye's line. Each number carries a ± range. That range comes from measuring your photo and three re-framed copies of it. On one of our own test selfies, the four readings were 7.01°, 7.81°, 7.31° and 7.46°, so the result showed about ±0.4°. That is the smallest range the tool will ever show, because claiming more precision than that would be false.

There is a second limit worth knowing. Google reports that its landmarks land, on average, 2.06% to 3.85% of the distance between the eyes away from where human annotators place them. On an adult face that is roughly 1.3 to 2.4 mm. An eye is about 30 mm wide (Packiriswamy et al. measured 30.91 mm in men and 29.62 mm in women), so a corner placed 1 mm too high or low moves the angle by about 2°. Differences of a degree or two between your eyes, or between you and a friend, are within that error.

The scan's quality check runs first. If your head is turned more than 8° it warns you, beyond 15° it asks for a retake, and it also flags chin up or down, tilt, smiling, closed eyes, low resolution and poor light. A smile matters here more than for most measurements because it lifts the lower lids and moves the outer corners.

You don't need to do anything about your canthal tilt. If you want to use it, these are the steps with sources behind them.