What Is My Face Shape? Measure the Pattern First
Quick answer: how to determine your face shape
To determine your face shape, use a straight-on photo or mirror view, pull your hair away from your forehead and jaw, and keep your head level. Record the widest forehead span, cheekbone span, full jaw span, hairline-to-chin length, and which horizontal line is widest. Then compare the ratios and the jaw outline instead of naming the shape from one selfie.
If you came here asking what is my face shape or what face shape do i have, keep the same measurements and treat the result as a working label rather than a one-selfie verdict.
Use cheekbone width as the reference W:
```text
length_ratio = length / W
forehead_ratio = forehead / W
jaw_ratio = jaw / W
```
Write the result as a working label plus a nearby label when the measurements overlap. Retake the photo from farther away under the same light before changing the label. This is the practical answer to both how to determine face shape and how to determine your face shape: measure the same landmarks twice, identify the width peak, and use the jaw outline only as a tie-breaker.
A repeatable face-shape decision record
| Pass | Record | Decision it supports |
|---|---|---|
| 1 | Photo distance, light, head angle, and hair position | Whether a changed result may come from the setup rather than your proportions |
| 2 | Forehead, cheekbones, full jaw, and length | The four readings used by the chart |
| 3 | length_ratio, forehead_ratio, and jaw_ratio | Which face-shape rows remain plausible |
| 4 | Width order from largest to smallest | Whether the forehead, cheekbones, or jaw is the first filter |
| 5 | Rounded, angular, pointed, or unknown jaw outline | Which close pair to compare next |
| 6 | Primary label, nearby label, and the reading that separates them | What to retest instead of forcing a single label |
If the second pass changes only one reading, keep both records and inspect that landmark. If several readings move together, repeat the setup first. A measured overlap is a useful result: it tells you what to check next and keeps the label from pretending to be a universal diagnosis.
| Measurement | Oval | Round | Square | Heart | Diamond | Oblong | Triangle |
|---|---|---|---|---|---|---|---|
| Forehead width (F/W) | <1.00* | <1.00* | >=0.82† | >=0.90† | <1.00* | <=1.00* | <1.00* |
| Cheekbone width (B/W) | <=1.00* | 1.00* | <=1.00* | <=1.00* | 1.00* | <=1.00* | <=1.00* |
| Jaw width (J/W) | <1.00* | <1.00* | >=0.82† | <=0.70† | <1.00* | <=1.00* | 1.00* |
| Face length (L) | 1.50W* | approx. 1.00W* | <=1.20W† | >1.00W* | >=1.25W† | >1.50W* | <=1.55W† |
| Length-to-width ratio (L/W) | 1.50x* | approx. 1.00x* | <=1.20x† | >1.00x* | >=1.25x† | >1.50x* | <=1.55x† |
W is the widest horizontal reading in the manual table. F/W is forehead divided by W, B/W is cheekbone divided by W, J/W is jaw divided by W, and L/W is face length divided by W. The cells marked with * are direct numeric anchors or inequalities from the measurement records. The cells marked with † are comparison bands used by this page to make close patterns inspectable; they are not universal diagnostic cutoffs. In the Triangle column, <=1.55x† is only the page's long-face comparison boundary, not a Triangle-specific standard.
Fill the Five Rows Without Changing the Method
Use one straight-on photo or one mirror view. Pull the hair away from the forehead and jaw, keep the head level, and mark the same five points every time: the widest forehead span, the widest cheekbone span, the jaw width, the hairline-to-chin length, and the widest horizontal line used as W.
Do not compare a one-sided jaw reading with a full-width reading. If you measure from the chin to one jaw turn, double it before comparing it with the forehead or cheekbones. Write the numbers in the same order before you name the shape.
One measurement record anchors an oval face around 1.50x face length and a longer face at 1.60x or more. A separate hands-on record uses 1.50x as the oval anchor and more than 1.50x as the long-face boundary. The same record can be written as W < 2/3 L for a long face. Those anchors are useful because they make a close oval-versus-oblong result inspectable instead of turning it into a visual guess.
The first pass should answer two questions:
- Which horizontal line is actually the widest?
- Is the lower outline rounded, angular, or wider than the cheekbone line?
The ratio narrows the field. The jaw and the width peak finish the decision.
Treat a Photo as a Measurement, Not a Verdict
A front-facing camera can make a face look wider or longer when the phone is tilted, held too close, or using a wide-angle lens. Take the second photo from farther away, keep the lens at eye level, and compare the numbers rather than comparing the two silhouettes by memory.
If the two photos produce different labels, keep the readings. A result such as oval with oblong length is more useful than throwing away the measurement that created the disagreement. Retest before changing your hair, glasses, or makeup to chase a label.
One tape-measure record came from a person with blond hair. They held the tape straight just in front of each facial area without letting it touch the skin, and the result came out oval. They had just cut their hair to match the oval example and kept the styling choice that emphasized the eyes while leaving the lips neutral.
Another person with thick hair and small features tried the measurements but found that the numbers did not make sense. They kept the oval label as a working result and reported that short hair worked well with their features.
A separate record used a 21 cm face length and a 12 cm width. Measuring from chin to hairline and across the widest point put the person between oval and long; they chose oval as the closer description.
Use a Frame as a Stress Test
A real object can expose a weak face-shape guess faster than another filtered selfie. The practical test is to keep the face and lighting stable, then change one frame variable at a time.
One try-on used a comfortable pair with a 138 mm front as the control. A TAPOLE Conan 113 frame was about 5 mm wider. The frame itself weighed 22.3 g, and the finished pair was about 35 g. The wider front changed the visual balance and made the hair and face read broader, so the useful correction was to question width before declaring the face shape wrong.
Another route started with a 30-minute 3D face scan in a store. The first set of frames did not solve the fit problem, so the test moved online to a Levi's frame. The frame was listed at 9 g, the lens choice was 1.60 blue-light aspheric, and the recorded total was ¥528. The point was not that one brand belongs to one face shape. The point was that a failed in-store route led to a more measurable second test.
A third comparison moved between BJ7139-B12 and BJ7180-B12 before the lens choice changed to ZEISS 1.60. The recorded calculation was ¥599 + ¥270 + ¥18 = ¥887. Keep that kind of record because it separates a face-shape question from a lens, bridge, or budget problem.
These tests are useful because they preserve the failed attempt. If pair A is too wide, do not change the frame family, the lens depth, and the color at the same time. Change one variable, take another straight-on photo, and keep the measurement that survived.
Read the Border Cases in the Right Order
Oval versus oblong: start with L/W. A manual result near 1.50x can read as oval in one chart, while a result at or above 1.60x is a stronger oblong signal. If the length sits between those anchors, inspect the jaw and the width distribution instead of forcing a single word.
Round versus square: both can sit near a 1.00-1.20 length-to-width band. Use the jaw row next. A value below 0.82 leans away from the square band used here; a jaw at or above 0.82 keeps square in the running. The number does not replace the visible curve or corner, but it tells you why the comparison is close.
Heart versus diamond: look at the width peak. The heart comparison band keeps the forehead at or above 0.90 W while the jaw is at or below 0.70 W. The diamond comparison band keeps the forehead at or below 0.78 W and the jaw at or below 0.74 W. The chin alone cannot settle this pair.
Triangle versus the upper-face shapes: a triangle reading needs a lower-heavy map. The jaw reaches 1.00 W as the widest horizontal line, while the forehead stays at or below the comparison band marked in the table. If the jaw does not remain the widest line after a second photo, keep triangle as a hypothesis rather than a final answer.
One oblong-face record paired a bob with darker powder and contouring. The person said the bob worked, and that the contour made the face look fuller.
A first-time glasses wearer with -1.75 in each eye and no astigmatism first obtained an in-person refraction and pupillary distance, then ordered a 1.56 lens online. After a week, they reported no dizziness, visual distortion, edge aberration, or all-day eye strain, and said they would buy the setup again.
Change One Variable After a Failed Test
Use this order when the first answer or first fit feels wrong:
- Retake the straight-on photo without changing the measurement landmarks.
- Recalculate
F/W,J/W, andL/Wfrom the same photo. - If the length ratio is stable but the fit is not, test frame width or bridge size.
- If the width peak changes, check camera angle, hair, and jaw measurement before changing the shape label.
- Keep a primary label and a secondary tendency when the result remains within one boundary.
This is the useful difference between I look round today and my face reads short, my jaw is below 0.82 W, and the frame that fit my square-faced friend was too wide on me. The second sentence tells you what to test next.
Answer the Question in One Sentence
Once the numbers are stable, answer what is my face shape? with the evidence attached:
My face reads as [primary shape] with [secondary tendency], because F/W is [number], J/W is [number], L/W is [number], and the widest line is [forehead, cheekbone, or jaw].
That answer stays useful when a different chart uses a different name. The label is only the short version. The five rows explain why it fits, where it is close, and what to test next.