Locate the eye corners
The model estimates the medial and lateral canthi on both eyes. Their normalized coordinates are converted into image pixels so horizontal and vertical distances use the same scale, even in a rectangular photo.
Use this free canthal tilt calculator to measure your eye tilt angle in degrees. Choose a level, front-facing photo to locate your inner and outer eye corners, correct minor sideways head roll and see your left, right and average readings.
Free · No signup · Your photo never leaves your device
Drag and drop or choose a file
JPG, PNG up to 10MBThe guide is measured from this sample photo.
Generated sample · measured with this face tool
Canthal tilt, called palpebral fissure inclination in clinical anthropometry, is the angle between a horizontal reference and the line joining the corners of an eye. The medial canthus is the inner corner near the nose; the lateral canthus is the outer corner near the temple.
The sign describes whether that line rises or falls toward the temple. Eyelid contour, brows and expression also shape how the eye area looks. For the wider facial outline, use the face shape detector.
MediaPipe Face Landmarker maps 478 facial landmarks in your browser. The calculation uses four eye-corner landmarks and an eye-line reference to measure each side separately.
The model estimates the medial and lateral canthi on both eyes. Their normalized coordinates are converted into image pixels so horizontal and vertical distances use the same scale, even in a rectangular photo.
For each eye, the code averages its inner corner, outer corner, upper-lid and lower-lid points to estimate a center. It rotates the coordinates until the line between these centers is horizontal, correcting sideways head roll. These are landmark-based centers, rather than detected pupil centers.
For each eye: θ = atan2(y_inner − y_outer, |x_outer − x_inner|) × 180 / π. Image y-coordinates increase downward, so this convention gives a positive angle when the outer corner sits higher.
Each eye angle is rounded to one decimal place. The displayed average is calculated from those two readings and rounded again; its direction determines the overall label. Decimal display precision does not establish measurement accuracy.
Use the table to interpret the average shown in your result. The neutral buffer is a display convention around zero; it is not a clinical reference interval or a measured error margin.
| Measured range | Direction | Geometric description |
|---|---|---|
| Above +1° | Positive | The outer corner is higher than the inner corner. |
| −1° to +1° (inclusive) | Neutral | The corners are close to level. |
| Below −1° | Negative | The outer corner is lower than the inner corner. A downturned eye line is a natural anatomical variation. |
For population context, Vasanthakumar et al. (2013) measured standardized frontal photographs of 200 South Indian adults aged 18–26: 100 men and 100 women. Mean inclination was +5.05° in men (SD 2.47°) and +6.10° in women (SD 2.85°). These are sample means with standard deviations, not a universal normal range. The study used a horizontal reference through the medial canthi; this calculator levels an estimated eye-center line, so the methods are not identical. See the clinical references below.
Use your result as a starting point for a look you want to try. Makeup and frame shapes change the visual line around the eyes; fit, eyelid shape and personal preference matter more than a category.
Try eyeliner that follows the existing outer-eye direction. A shorter, softer wing may suit you if a steep wing feels too sharp. Rounded or rectangular frames offer different amounts of contrast with an upturned eye line.
Keep liner close to the lash line for a straighter look, or angle the outer portion upward for a cat-eye effect. Try both rounded and angular frames, checking how the upper rim sits against your brows.
You can follow the natural downturned contour or try a lifted finish. For a lift, begin a short wing before the outer end of the upper lash line rather than tracing the corner downward. Subtle cat-eye or browline frames with a rising upper rim are another option to compare.
Head pitch (chin up or down), yaw (turning left or right) and close camera distance change the eye-corner line projected into a 2D photo. Eye-line correction handles sideways roll, but cannot reconstruct the missing 3D perspective.
Keep both sides equally visible, look straight ahead and avoid raising or tucking your chin. A head turn compresses one eye in the photo.
Place the lens near eye level and at least an arm’s length away. Moving the camera farther away reduces close-up perspective distortion; zooming alone does not change perspective.
Light both eye corners evenly and relax your eyes. Squinting and strong shadows make the visible eyelid junction harder to locate.
Remove glasses and move hair away. Avoid filters and heavy liner over the corners. Use the same distance, pose and lighting for repeat measurements.
The studies below cover population measurements, age differences and photographic pose. The numerical example above comes from the first paper.
Vasanthakumar, Kumar & Rao (2013), Oman Medical Journal — Anthropometric Analysis of Palpebral Fissure Dimensions and its Position in South Indian Ethnic Adults. Table 1 reports the means and standard deviations quoted above.
Park et al. (2008), Plastic and Reconstructive Surgery — Anthropometry of Asian Eyelids by Age. Reports eyelid measurements across age groups and illustrates why reference values depend on the population studied.
Vaca et al., ASJ Open Forum — The Influence of Sagittal Head Tilt on Periorbital Appearance. Standardized photographs of 12 female volunteers showed more pronounced positive tilt with downward head pitch.
Bashour (2006), Plastic and Reconstructive Surgery — History and Current Concepts in the Analysis of Facial Attractiveness. Reviews several contributors to facial perception, providing context beyond an isolated eye angle.
Quick answers before you test another photo.
Natural asymmetry, eyelid position and expression can produce different readings. A slight head turn or an obscured corner can widen the gap. Retake the photo with both sides equally visible; the facial symmetry test compares more paired landmarks.
No. The browser downloads the detection model, then runs landmark detection and angle calculations on your device using JavaScript and WebAssembly. The selected image is not sent to our server. Use Clear photo to remove it from the workspace.
There is no single average that applies to everyone. The South Indian young-adult sample in the ranges section provides one measured example; other studies use different populations, ages and reference lines. Compare the study method as well as the number. A sample mean is not an ideal angle or a cutoff for an individual.
It can. Changes in eyelid skin and supporting tissues can alter the visible eye contour and corner position over time. The amount and direction vary between people. Comparing older and newer photos is more useful when pose, expression and camera setup match.
Yes. A liner wing or frame rim creates an extra visual line around the eye. It can also obscure the actual corner and shift automated landmark placement. Measure with glasses removed and the corners visible, then compare styling choices using the guide above.
Tilting the chin downward can make positive tilt appear more pronounced; raising it can reduce that appearance. Vaca et al. observed this pose effect in standardized photographs. Camera height changes the viewing angle too, but there is no fixed degree correction for a casual selfie. Keep the lens at eye level and your chin level.
Canthoplasty and canthopexy can alter or support the outer eyelid corner in selected cases. Suitability, risks and likely results require assessment by a qualified oculoplastic surgeon. This photo measurement cannot determine whether a procedure is appropriate.