Round opening
More visible vertical openness relative to corner-to-corner width.
Free Eye Shape Detector:What Eye Shape Do I Have?
Measure round, almond, or elongated opening geometry and a separate upturned, level, or downturned corner direction. Lid-fold appearance is available as an optional visual self-check rather than an automated claim.
Upload one centered face photo
Use a level front photo with both eyes relaxed and naturally open so their visible outlines can be compared.
Drag and drop your photo here
or choose a file below
JPG, PNG, WebP or HEIC · Up to 30 MB
Best results: eye-level camera, direct gaze, relaxed eyes, no glasses.
🔒 Private: core analysis stays on your device. Your photo is shared only if you separately opt in to an eligible AI photo feature.
Example framing only, not a target look.
How to take a good photo
For the best results:
- Face straight on (not turned)
- Natural lighting
- Face centered in frame
- Remove glasses or hair covering forehead
- Relaxed expression
Results are based on visible eyelid margins, eye corners, spacing, and bilateral agreement.
Need more details? See FAQ
How the Eye Shape Detector Works
The detector measures the visible opening of each eye from its inner and outer corners and upper and lower eyelid margins. Width is divided by perpendicular aperture height, so sideways image roll does not create a false opening shape.
Opening shape and corner orientation are reported independently. Lid-fold appearance is not inferred from landmarks; after the automatic result, you can optionally compare your relaxed eyes with the visual reference and add a self-reported description.
Trace
Map both visible eyelid margins and the inner and outer corners.
Measure
Calculate each width-to-height ratio and check bilateral agreement.
Describe
Combine the automatic measurements and optionally add your own visual lid-fold description.
Attribution does not grant reuse rights. Read the image reuse terms.
Text descriptions of the eye-shape reference
Almond opening
A balanced, tapered opening between the round and elongated bands.
Elongated opening
More horizontal width relative to the visible upper-to-lower aperture.
Downturned corners
The outer corner sits lower than the inner corner after roll correction.
Approximately level corners
The inner and outer corners sit near the same corrected horizontal level.
Upturned corners
The outer corner sits higher than the inner corner after roll correction.
Visible crease
A distinct upper-lid crease remains visible above the lashes.
Hooded appearance
The upper-lid fold sits close to the lashes and reduces visible lid space.
Monolid appearance
No distinct crease is visible in the sampled upper-lid region.
How Opening Shape Is Classified
The product bands use width divided by perpendicular lid aperture. They are transparent descriptive labels, not clinical norms or attractiveness targets. Results close to either boundary say so explicitly.
Round
< 2.6×More visible vertical aperture relative to corner-to-corner width.
Almond
2.6–3.3×An opening proportion between the round and elongated bands.
Elongated
> 3.3×More horizontal width relative to the visible lid aperture.
Three Traits That Should Not Be Collapsed Into One
| Trait | Possible descriptions | Evidence | Tool status |
|---|---|---|---|
| Opening shape | Round · Almond · Elongated | Eye-corner width and perpendicular upper-to-lower lid aperture | Measured |
| Corner direction | Downturned · Approximately level · Upturned | Inner-to-outer corner angle after detected roll correction | Measured |
| Lid fold | Visible crease · Hooded · Monolid | Your visual comparison with relaxed eyes | Optional · self-reported |
Get a More Reliable Result
Keep the camera at eye level with both sides of the face equally visible.
Look directly at the lens with relaxed, naturally open eyes.
Remove glasses and keep hair, glare, and heavy shadow away from the eyelids.
Use a neutral expression; squinting, smiling, or widened eyes change the aperture.
Head turn lowers result confidence; genuinely unstable bilateral evidence produces a retake request without consuming the free result.
Accuracy, Privacy, and Scope
Core landmark analysis runs in your browser. Your photo is shared only if you explicitly opt into a separate AI styling preview.
The main label checks both eyes. It does not hide large left-right disagreement inside a simple average.
This is visible two-dimensional geometry from one photo—not a diagnosis, identity classifier, or attractiveness grade.
Frequently Asked Questions
What eye shape do I have?
Eye shape is easier to describe on separate axes. The visible opening can be round, almond, or elongated, while the corner direction can be downturned, approximately level, or upturned. This detector measures both from a suitable front photo.
How does the eye shape detector work?
It traces the visible eyelid margins of both eyes, divides corner-to-corner width by perpendicular lid aperture, checks left-right agreement, and measures corner direction after detected head roll is corrected.
Can an eye be both almond and upturned?
Yes. Almond describes the visible opening proportion, while upturned describes the corner direction. Because they are independent traits, they can appear together in one result.
Does the detector identify hooded eyes or monolids?
No. Face landmarks trace the visible lid margins but not the upper-lid crease or overlying fold reliably. The result therefore offers an optional visual self-check for visible crease, hooded, monolid, or not sure instead of forcing an automated label.
Why are my left and right eye measurements different?
Some bilateral difference is normal. Gaze, squinting, expression, head turn, lighting, glasses, or a partly hidden eyelid margin can also increase the difference in a photo. Strong disagreement produces a retake request rather than a forced label.
Can makeup change my measured eye shape?
Makeup can change visual emphasis but not the underlying anatomy. Heavy liner, lashes, or shadow can also hide eyelid margins and make automated measurement less reliable.
Is one eye shape more attractive?
No opening or corner-direction category is inherently better. The labels describe photographed geometry rather than attractiveness, health, ethnicity, identity, or personal value.
Result safeguards
What the detector does—and what it does not do
Runs on your device
Core landmark analysis happens in this browser.
Both eyes measured
Openings and corner directions are checked on both sides.
Unstable photos get a retake
Pose lowers confidence; unstable bilateral evidence triggers a retake.
Descriptive, not evaluative
No beauty, health, ethnicity, or identity ranking.