Measure the points you define

Eye Symmetry Test

One eye can look higher without being wider or more tilted. Mark four eye corners to compare these three relationships separately in your photo. Keep your head level: the measurements use the photograph’s horizontal axis.

Eye symmetry toolPrivate · Manual points

Choose a photo to place reference points.

Measure eye symmetry in your photo

Mark defined points on your photo. Read the geometry as you adjust them.

Tap the photo to place the selected point. Arrow keys move a crosshair; Enter or Space places it. Shift moves faster. Exact coordinates are available beside the photo.

Dashed blue lines show photo-horizontal references and the midpoint-height comparison. Hollow diamonds mark corner midpoints; green circles are your selected points.

JPG, PNG or WebP · Up to 20 MB · Processed only in this browser.

Place defined reference points

Mark the four visible eye corners. Angles use the photograph’s horizontal axis; head roll changes the apparent height difference.

Enter exact coordinates

Coordinates are percentages from the image’s left and top edges.

Your photo measurements

Place all reference points to see the measurements.

Defined manual geometry. No score or population ranking.

Values normally show two decimals; very small nonzero values use scientific notation (for example, 1e-3 means 0.001). The JSON keeps the computed values. These measurements describe a 2D photo. Perspective, expression, head roll and your point selection affect the result.

Start with the same landmarks

Four corners, three different questions

Mark where the upper and lower eyelid margins meet. The inner corner (medial canthus) is toward the nose; the outer corner (lateral canthus) is toward the temple. Follow the actual meeting point, rather than a lash tip, eyeliner wing or eyelid crease.

Original schematic with P1 outer and P2 inner on the image-left eye, P3 inner and P4 outer on the image-right eye. M1 and M2 are the midpoints of the corner lines, not the pupils. The lines compare width, tilt and midpoint height.
The numbered corners enter the calculation. M1 and M2 are calculated halfway along each corner line; they are not detected pupil centers. Eye outlines are illustrative.

“Image-left” means the left of the displayed photo. A mirrored selfie may swap which anatomical side appears there. Use the displayed point labels throughout.

Read each measurement on its own

What the results actually tell you

Corner tilt

Each signed angle compares the corner line with photo horizontal. Positive means the outer corner is higher; negative means it is lower. The third angle value is the absolute difference between the two tilts.

Midpoint height

The eye-center height difference is the vertical gap between M1 and M2, divided by their straight-line spacing, times 100. It is an absolute percentage, so it does not identify the higher side. It does not establish eye-socket alignment.

Width ratio

Eye-width ratio is image-left corner distance divided by image-right corner distance. 1.10 : 1 means the left line is 10% longer relative to the right line. 1.00 : 1 means those lengths match; it says nothing about eyelid opening.

These values have no combined symmetry grade, validated normal range or attractiveness ranking. Matching widths can coexist with different heights or tilts. A ratio of 1.00 is not evidence that every visible eye feature matches.

Make the comparison fair

Choose a photo you can measure consistently

Face the camera straight on at eye level, keep your chin neutral and relax your expression. Use even light with both corners visible. A blink, squint, sideways turn, glare or shadow can make a corner harder to place. Choose a clearer image instead of guessing through an obstruction.

Keep distance and framing consistent across retakes. Move the camera farther away and crop afterward if a close selfie changes the apparent proportions. The selfie perspective guide explains that issue.

  1. Choose a photo or try the sample.JPG, PNG or WebP, up to 20 MB. Photo processing stays in this browser.
  2. Place P1, P2, P3, then P4.Outer, inner, inner, outer from image-left to image-right. Select a point again to refine it.
  3. Check the points before the numbers.Arrow keys and Enter or Space can place points; exact percentage coordinates are also available.
  4. Keep a useful record.Save the measurement JSON or marked PNG with a note of pose and expression. The PNG includes your photo.
An example you can check

Equal tilt does not mean equal width or height

Consider these illustrative working-image pixel coordinates: P1 (100, 200), P2 (232, 200), P3 (346, 212), P4 (466, 212). Both corner lines are horizontal, so both tilts and their difference are 0°.

Compare corner distances

width = √(Δx² + Δy²)

The widths are 132 and 120 pixels. Their ratio is 132 ÷ 120 = 1.10 : 1. Width uses straight-line distance, including any vertical component; it is not just the horizontal span.

Compare corner midpoints

height gap (%) = 100 × |M2y − M1y| / distance(M1, M2)

M1 is (166, 200) and M2 is (406, 212). Their spacing is √(240² + 12²) ≈ 240.30 pixels. The result is 100 × 12 ÷ 240.30 ≈ 4.99%, rather than 12%.

For each tilt, the calculator uses atan2(inner y − outer y, |outer x − inner x|) × 180 / π. Image y increases downward, so a higher outer point produces a positive angle.

Uniform resizing preserves all five results. Stretching the photo can change them. Calculate distances in working-image pixels: x and y percentages represent different pixel lengths in a non-square image.

The photo has a reference frame

A level-looking eye pair can gain a height gap when rotated

Start with matching horizontal eye lines and midpoints at the same height. Rotate that flat drawing clockwise by 3°: the width ratio stays 1.00, but the midpoint-height result becomes 100 × sin(3°) ≈ 5.23%. The image-left tilt becomes +3° and image-right becomes −3°, producing a 6° tilt difference.

Same schematic before and after clockwise image rotation of three degrees. Before: width ratio 1.00, midpoint gap zero and both tilts zero. After: width ratio 1.00, midpoint gap 5.23 percent and tilts plus three and minus three degrees.
Rigid rotation changes the relationship to the photo edge. This is an exact geometric illustration, not a correction applied by the test.

Chin-up or chin-down pose introduces a different perspective effect. Vaca et al.’s study of 12 female volunteers found that sagittal head tilt changed photographed eyelid and canthal measurements. Its findings support consistent photography; they do not supply a universal correction.

To check placement sensitivity, reset the points and mark the same photo once more. A large change tells you the endpoints were difficult to repeat. Agreement only checks your repeatability on that image; it does not prove anatomical accuracy.

Common questions

When your eyes look different in a photo

Why can one eye look smaller when the width ratio is 1.00?

The visible eyelid opening, crease, brow position and surrounding shadows are separate from corner-to-corner width. This four-point test does not trace the upper and lower lids, measure iris diameter or identify which structure explains the appearance.

Will flipping the image change my result?

A horizontal flip swaps image-left and image-right. With corners correctly relabeled, the width ratio becomes its reciprocal: 1.10 becomes about 0.91. The two tilt values swap; their absolute difference and the midpoint-height percentage stay the same. A flip changes orientation, not the original geometry.

Does a higher percentage mean I have a medical eye problem?

No. This percentage is a photo-reference measurement without a diagnostic threshold. It cannot diagnose ptosis, strabismus or orbital differences. The American Academy of Ophthalmology defines ptosis as upper-eyelid drooping, which requires an eye examination to assess; four corners do not measure it.

What if the difference is new or my vision changes?

Seek professional assessment rather than relying on another photo measurement. NHS guidance on double vision recommends urgent help when double vision starts suddenly, and emergency care for double vision after a head injury or severe headache with blurred or double vision. A photo test cannot assess these symptoms.

Is this an AI eye test, and can it show treatment progress?

You place all four points manually; no detector or automatic leveling runs here. Different expressions, poses and point choices can change the readings. A repeated photo measurement cannot establish a treatment effect. Follow a clinician’s photography instructions when a medical comparison is needed.

Sources and method notes

Research checked October 17, 2026. The diagrams and worked calculations are original. This guide does not claim clinical validation or medical review.

For corrections to these instructions, contact the site.

Privacy settings

Photo analysis works without analytics. You can optionally allow Google Analytics to measure basic page visits. We do not send photos, filenames, face measurements or scores to analytics.

Analytics is off unless you allow it.

Your choice is stored in this browser. Turning analytics off after allowing it reloads this page; save any wanted result first. Read the privacy policy.