How the canthal tilt test works
- 1
Add a front-facing photo
Choose a clear, front-facing photo taken at eye level with a relaxed expression. Before any angle is calculated, an automatic quality check rejects photos with no face, more than one face, a face that is too small, a photo that is too dark, or a head turned or tilted too far.
- 2
Place the 3D face mesh
A face landmarker model runs locally in your browser tab and places 478 three-dimensional points across your face, including the inner and outer corners of both eyes. Because the model runs on your device, your photo never leaves your phone or computer.
- 3
Measure both eye angles
The tool draws a reference line through your two inner eye corners, which rotates with your face and so removes head roll. For your right and left eye separately, it measures the angle from the inner corner to the outer corner against that line, in degrees.
What we measure
Canthal tilt is the upward or downward slope of the eye opening from the inner corner near the nose to the outer corner near the temple. Alongside the tilt angle, the tool reports three related eye and brow measurements.
Canthal tilt
DeviceThe angle in degrees of the line from the inner corner of the eye to the outer corner, measured against the reference line through both inner corners. Because that line rotates with your face, a slight head roll does not skew the result. Values are reported separately for the right and left eye, and a positive number means the outer corner sits higher than the inner corner.
Eye height to width
DeviceA ratio of how open the eye is vertically compared with its width. For each eye, the tool averages two gaps between the upper and lower lid and divides by the corner-to-corner eye width. On our calibration faces, the middle 80% ranged from 0.207 to 0.357, with a median of 0.294.
Eye shape
DeviceA plain-language label for your eye shape, such as Almond, Round, Upturned, Downturned or Narrow. For the tilt part of the label, an average tilt above +7.7° across both eyes is read as clearly upturned, and an average below +1.3° is labeled downturned.
Brow tilt
DeviceThe angle from the inner brow head to the outer brow tail, measured against the across axis of the face. On the 3D face mesh this angle is usually negative. On our calibration faces the median was about −7.9°, and the middle 80% ranged from −12.6° to −3.0°.
What canthal tilt is and how people use it
Canthal tilt describes the slant of the eye opening, the gap between the upper and lower eyelids. It compares the outer corner of the eye (lateral canthus) with the inner corner near the bridge of the nose (medial canthus). When a line from the inner corner to the outer corner angles upward toward the temple, the tilt is positive. When the line is level, the tilt is neutral, and when the outer corner sits lower than the inner corner, the tilt is negative. On this tool, readings are grouped into the bands shown below, so the low band covers neutral and negative readings.
Canthal tilt is one way to describe the geometry of the upper face. Because the tool reports an angle for each eye, you can compare your left and right eye, or see how your eye corners sit before you choose brow shapes or glasses frames.
Positive tilt
Level corners
Negative tilt
Each eye is drawn from inner corner (left) to outer corner (right). Pink: corner line; dashed: face horizontal.
How our tool measures canthal tilt step by step
When you select a photo, the tool loads Google's MediaPipe Face Landmarker inside your browser tab. The model file is about 4 MB and is downloaded once. All processing happens on your device, so the photograph is never uploaded to a server. The calculation is deterministic, so the same photo always gives the same degree measurements.
Before placing any measurement lines, the tool runs a quality check. It rejects photos with no face, more than one face, a face that is too small, or a photo that is too dark. It also rejects photos where the head is turned too far to the side (yaw) or tilted too far up or down (pitch).

Pink dots mark detected facial features; pink lines connect the measured widths, face length and eye corners.
Next, the model estimates 478 three-dimensional landmarks across the face. The MediaPipe documentation describes this output as an estimate of 478 3D face landmarks. An earlier published mesh model with 468 vertices, described by Kartynnik and colleagues (2019), was reported to reach prediction quality comparable to the variance in manual annotations of the same image. From the mesh, the tool takes the inner and outer corner points of both eyes.1
Finally, the tool measures the angle between the reference line through your two inner eye corners and the line from each inner corner to its outer corner. Because the reference line rotates with your face, in-plane head roll is removed. The tool shows the angles in degrees for both eyes, draws the measurement lines on your photo, and lets you save a share card with the photo turned off by default.
What published research says about eyelid and brow measurements
Published anthropometric studies measure eyelid and eyebrow dimensions in different groups. Kunjur, Sabesan, and Ilankovan (2006) studied the position of the eyebrow and eyelids in white, Indian, and Chinese groups. They note that these measurements differ among ethnic groups, and their aim was to quantify the positions and test whether the groups differ significantly.3
Rhee, Woo, and Kwon (2012) used photogrammetric measurements on seven composite faces of different races or ethnicities. They took fourteen distance measurements and five angular measurements to compare the eye configuration of average and attractive composite faces. The authors report some common and some different characteristics across groups, including measurements of the palpebral slant, eye fissure height and width, and brow position.2
These studies used different groups and methods, and they measured people or composites directly rather than using a 3D face mesh. For that reason, this page does not compare you with any group. It only reports your angles and shows where they sit against our own calibration faces.
How to read your number against our calibration band
The landmarker fits a canonical 3D mesh to a 2D photograph, so real faces span a narrow band of ratios and angles on the mesh. Mesh angles are not the same as measurements taken directly on a person. To help you interpret your numbers on this tool, we compare them against 133 real front-facing photos from Pexels. These calibration photos help us interpret measurements made with this mesh. They are a mesh reference, not population norms.
On our calibration faces, the middle 80% of left-eye canthal tilt measurements ranged from +1.0° to +8.7°, with a median of +4.8°. The bands on this page are: above +7.7° is read as clearly upturned, +1.3° to +7.7° is positive, and below +1.3° is low. The eye shape detector applies these cut-offs to the average tilt of both eyes and labels the low band as downturned.
Number of photos
- Median:
- 4.8°
- Middle 80%:
- 0.7°–8.5°
- 1Downturned < 1.3°
- 2Upturned > 7.7°
Your right and left eye can give different readings, and the tool reports each one separately. Small differences can come from the face itself, from expression, or from head position in the photo. If the two numbers are far apart, try another photo with a relaxed face and the camera at eye level before drawing conclusions.
Styling and photographing with your eye angle in mind
Styling changes how the eye area looks, not the measured angle, which comes from the position of your eye corners in the photo. Brow shaping and hair are two simple things to adjust. Keeping the outer tail of the brow tidy and keeping hair off the temples leaves the eye corners and brow ends visible in photos.
Glasses frames and photo setup are two other options. Frames with an upper rim that sweeps upward toward the temples can echo a positive tilt, while straight horizontal top bars give a more level look. For photos, keeping the camera at eye level and the chin level keeps the eye corners lined up and gives the cleanest measurement.
Reading your canthal tilt
On our calibration faces, left-eye tilt in the middle 80% ran from +1.0° to +8.7°. The three bands below describe our calibration photos and this face mesh, not population norms, and published eyelid studies vary by group and method.
| Tilt | Description | On our calibration faces |
|---|---|---|
| Above +7.7° | Clearly upturned | Upper part of our calibration range; the 90th percentile is +8.7° |
| +1.3° to +7.7° | Positive | Includes the calibration median (+4.8°) |
| Below +1.3° | Low, labeled Downturned | Lower part of our calibration range; the middle 80% spans +1.0° to +8.7° |
What the canthal tilt test can't tell you
- Expressions such as smiling or squinting can shift the eye corners, so a relaxed face gives the cleanest reading.
- The corner points come from an automated 478-point face mesh, so they can be off on low-resolution or blurry photos.
- Close-up wide-angle selfies and a head turned or tilted up or down can affect the landmarks. The tool rejects photos with too much yaw or pitch, but smaller amounts can still matter.
- Mesh angles are not identical to protractor or caliper measurements taken in person, so numbers from this tool should not be compared directly with published studies.
- Our reference bands come from 133 real front-facing photos used as a mesh reference, not from a population study.
- Canthal tilt is a single geometric angle of the eye opening. It is not a measure of attractiveness, health, or personality.
Photo tips
Keep the camera at eye level
Hold your phone or camera directly in front of your face at eye height and look straight into the lens. Tilting your chin up or down can change how the eye corners line up, and the quality check rejects photos tilted too far.
Use soft, even front lighting
Stand facing a window or a soft light source so both eyes are lit evenly. The quality check rejects photos that are too dark, and even light helps the landmarks sit on the eye corners.
Keep a relaxed, neutral expression
Relax your forehead, cheeks, and eyelids before taking the photo. Smiling or squinting can move the eye corners and change the gap between your lids.
Pull hair away from your temples and brows
Tuck bangs, fringes, and loose strands behind your ears so your brow tails and outer eye corners are fully visible.
Questions
References
- Kartynnik, Y., Ablavatski, A., Grishchenko, I., & Grundmann, M. (2019). Real-time facial surface geometry from monocular video on mobile GPUs. arXiv:1907.06724. arxiv.org/abs/1907.06724
- Rhee, S. C., Woo, K.-S., & Kwon, B. (2012). Biometric study of eyelid shape and dimensions of different races with references to beauty. Aesthetic Plastic Surgery, 36(5), 1236–1245. doi.org/10.1007/s00266-012-9937-7
- Kunjur, J., Sabesan, T., & Ilankovan, V. (2006). Anthropometric analysis of eyebrows and eyelids: An inter-racial study. British Journal of Oral and Maxillofacial Surgery, 44(2), 89–93. doi.org/10.1016/j.bjoms.2005.03.020