Positive Canthal Tilt: Do You Have One, and Is It Rare?

Canthal tilt is the angle of the line from the inner corner of the eye (the medial canthus) to the outer corner (the lateral canthus), compared with horizontal. If the outer corner sits higher, the angle is positive and the eye reads as upturned. Level corners are neutral. A lower outer corner is negative, which beauty guides call a downturned eye. Clinique's eye-shape guide describes upturned eyes in exactly these terms: inner corners that dip and outer corners that lift.

There is no official cut-off between “neutral” and “positive”. Researchers simply report the angle in degrees. FaceCheckup's style plan needs a working line, so it treats anything above +2° as positive, anything below −2° as negative and the band in between as neutral. That ±2° band is deliberately wide, because a corner point placed a millimetre off moves the angle by about two degrees on an adult eye (more on that below).

Two words you'll see in looksmaxxing threads, hunter eyes and fox eyes, are not measurements at all. They are informal labels that bundle a positive tilt with other things, such as a low brow, little visible upper lid or a long eye. Nothing in the research we use defines them, so this page sticks to the angle you can actually measure. If you want the full picture of tilt in both directions, the canthal tilt guide covers it.

No. The chart below plots every population average we could trace to a paper we read. The lowest is 1.2°, in a sample of Indian women quoted by Packiriswamy and colleagues (2012). The highest is 9.64°, also an Indian sample, quoted in the same paper from Hanada's work. The two largest studies we read in full put men and women at 4.71° and 7.05° (300 Malaysian South Indian adults, Packiriswamy et al. 2012) and at 8.5° and 8.8° (996 Korean adults, Choi and Park 2002).

So if your scan shows a positive tilt, you are in line with the average of every group in that list. What the papers don't publish is the share of people above or below zero, so any “only 15% of people have a positive canthal tilt” figure you've seen has no source we could find. The honest version is: averages are positive everywhere they have been measured, and individuals spread around them by roughly 2° to 3° (one standard deviation) in the large studies.

Most of the “do I have positive or negative tilt?” posts on looksmax.org end the same way: someone points out that the photo was taken from below, or with the head tipped, and the answer changes. That is the real problem with eyeballing tilt. The outer corner sits further back on the head than the inner corner, so tipping your chin down raises the outer corner in the picture and makes a tilt look more positive.

When we rotated Google's MediaPipe canonical face model and measured it the way a basic calculator does, a 10° chin-down tip added about 2.5° of tilt and a 10° chin-up tip took about 2.5° away. That is a geometry result for one template face, set out in full on the canthal tilt page, but it shows why a selfie taken from slightly above can manufacture a positive tilt.

FaceCheckup rotates your landmarks so the line between your pupils is level, and uses the 3D landmarks to bring a slightly turned or tipped head back toward frontal, before it measures each eye. You still get the best reading from a good photo:

The claim that a positive tilt is attractive usually traces back to one study. Bashour and Geist (2007) showed people pairs of female faces, one unmodified and one edited to accentuate the tilt, and the edited face was picked 93% of the time. That is a strong result, but it is a forced choice between two versions of the same faces, it used female faces only, and the abstract does not report how many judges took part.

A much larger study points the other way. Kempa and colleagues (2025, Journal of Plastic, Reconstructive and Aesthetic Surgery) had 2,081 participants, average age 26.5, rate edited versions of one unisex face on a five-point scale. Raising the outer eye corner lowered ratings of attractiveness, femininity, masculinity and trustworthiness, and raised ratings of dominance, threat and disgust. Raising the brows did much the same, and combining the two strengthened the effect.

The two designs differ, and neither one rated real, unedited faces from the general population, so neither settles the question. What they do show is that “more tilt is better” is not an established finding. Add the fact that private taste explains roughly as much of anyone's attractiveness judgements as shared taste does (Hönekopp 2006), and a few degrees of tilt starts to look like a small part of how a face is read.

Not by exercise, taping or mewing. The bone and tendon positions that set the eye corners don't change with facial exercises, and a 2025 systematic review found the evidence for mewing changing facial structure to be sparse and anecdotal. Surgical options exist, but they are a conversation for a qualified surgeon, not a website, and this site doesn't give surgical advice.

What does change the look of the eye is makeup and brows, and the sources we use agree on a few techniques. They work by drawing the eye upward, not by changing the corners themselves:

Tilt is only two points per eye. Several other features change whether eyes read as lifted, and the scan measures some of them. Eye openness (visible height ÷ width) affects how much the angle shows: a narrower eye with a positive tilt reads as more slanted than a round eye with the same angle. Brow height, measured from the upper lid to the top of the brow, changes the frame around the eye. And a tilt difference between left and right is normal: the result shows it separately, and a gap of a degree or two is inside the landmark error.

Lid shape matters too, but a landmark scan can't see it. A hooded lid, where skin folds over the movable lid, can make the outer part of the eye look heavier than its corners are. If your scan says positive but your eyes look downturned to you, a hood is a likely reason. The negative canthal tilt guide explains how to tell a low corner from a heavy lid.

Finally, the precision limit. Google's model card for the landmarker reports that points land, on average, 2.06% to 3.85% of the distance between the eyes away from where human annotators place them, which is roughly 1.3 to 2.4 mm on an adult face. On an eye about 30 mm wide, a corner a millimetre off moves the angle by about 2°. That is why a reading of +1° is better described as “level” than as “slightly positive”, and why we set the positive line at +2°.

The eyes section of your result shows canthal tilt averaged across both eyes with a ± range, the left-versus-right difference, eye openness and brow height. Next to the tilt you'll see the published span from the two large studies, group averages 4.7° to 8.8° and roughly 2° to 11° within one standard deviation, as context. The tilt is not part of the Balance Score; the how it works page lists what is.

If your number is above +2°, your style plan shows the positive-tilt liner rules. Between −2° and +2°, you'll see neither the positive nor the negative set, because the corners are level within the error. If one eye reads positive and the other negative, retake the photo first: head roll pushes one eye up and the other down by the same amount, and the level correction can only remove what the landmarks show.

Whatever the number, it describes an angle, not a verdict. The broader facial symmetry test and the face-shape tools on this site work the same way: measurements with their limits stated, and no ranking of you as a person.