What Is My Face Shape? A Measured Self-Test

Six labels cover nearly everyone: oval, round, square, heart, diamond and oblong. None of them describe a personality. Each is a claim about proportion, and the border between two neighbours is a number rather than a feeling. Read the table below as a set of tests you can run, not as a quiz result.

Most lists of face shape types stop at five and quietly drop oblong, which is why so many long faces get told they are oval. Keep oblong in. It is the shape that changes the haircut advice most, because adding height to a face that is already long is the one move that reliably backfires.

Borderline faces are normal. If two shapes are within a centimetre of each other on your tape, you sit on the border, and both sets of advice apply. That is a useful answer, not a failed test.

A soft tailoring tape and a mirror beat any quiz. The trick is repeatability: take each measurement the same way every time, write it down, and you never have to wonder again. Here is how to determine face shape in four passes.

1. 1

   Forehead width

   Lay the tape flat across the widest part of the forehead, roughly halfway between eyebrow and hairline. Keep it level. Curving it over the brow ridge adds a centimetre you did not earn.

2. 2

   Cheekbone width

   Measure from the outer corner of one eye, straight across the top of the cheekbones, to the outer corner of the other eye. This is the number people take too low, which turns a diamond into a square.

3. 3

   Jaw width

   Find the corner of the jaw below your ear, measure from there to the point of the chin, then double it. Running the tape under the jaw instead picks up your neck and inflates the reading.

4. 4

   Face length

   Centre of the hairline straight down to the base of the chin. If your hairline has moved, measure to where it used to sit and note that you did, because a receding hairline makes every face read oblong.

Now the two ratios. Length divided by cheekbone width near 1.0 puts you in round or square territory; near 1.5 puts you in oval, heart or diamond; well above that is oblong. Then compare forehead to jaw. Forehead clearly wider means heart. Jaw and forehead level with a corner means square. Cheekbones winning over both means diamond.

The photo route answers what is my face shape with no tape in the house, and it works if the photograph is honest. Face the camera square on, pull all hair back off the forehead and ears, hold a neutral expression with the jaw relaxed, and put the lens at eye level. A phone held low and close will widen the jaw and shorten the forehead, which is exactly the distortion that makes people answer what shape is my face incorrectly.

With a clean frontal shot you can run it through our face shape detector and compare the machine reading against your tape. Here is the whole run, as it appeared on 7 September 2026.

1. 1

   Open the tool

   The composer has three tabs, Generator, Template and Tools. The detector sits under Tools, and the helper line under the panel states what it does: analyse facial proportions and classify the face shape.

2. 2

   Load one photo

   Click the Photo box and drop in the frontal shot. One upload is all the tool takes. We used a plain-background headshot of a fictional woman so nothing in the frame competes with the face.

3. 3

   Set ratio and resolution

   The chips offer 16:9, 9:16, 1:1, 4:3 and 3:4 for shape and 1K, 2K or 4K for size. We left the defaults at 1:1 and 1K, because the output is a report card rather than a wallpaper.

4. 4

   Press Generate

   The button prints the price before you commit: 130 credits struck through, 13 credits actually charged. The job queues and the progress appears in My Image.

5. 5

   Swap the photo and repeat

   Replacing the file in the Photo box and pressing Generate again ran the second face for another 13 credits, so the full comparison in this article cost 26 credits.

The card that came back is more useful than a one-word verdict. It draws the four horizontal reference lines across the portrait, prints a conclusion, lists the visual features it used to get there, and shows a strip of reference silhouettes so you can argue with it. It also labels itself honestly at the top: a visual estimate, not a medical measurement. Every tool and template in the workbench runs on GPT Image 2, which is why the report is drawn rather than computed.

Two shortcuts get you close in under a minute. They are worse than the tape, and knowing how they fail is the point.

• The jaw and forehead comparison. Look straight into a mirror and decide which is wider, your forehead or your jaw, ignoring everything else. Forehead wider points at heart or oval; jaw as wide as the forehead points at square or oblong; neither obviously wider, with the cheeks winning, points at diamond or round.
• The outline trick. Stand close to a bathroom mirror, tie your hair back, and trace the edge of your reflected face on the glass with a dry-erase marker or a bar of soap. Step back and look at the shape you drew, not at yourself. This is the fastest way to see a corner on the jaw that you have been arguing about for years.
• The wall-shadow version. One lamp to the side, your profile against a pale wall, phone on a shelf. The cast outline reads length far better than a mirror does, because you cannot subconsciously tilt into a flattering angle.

Both mirror methods are drawing exercises, so they inherit your drawing errors, and both flatter. People trace a narrower jaw than they have. Treat them as a first pass that tells you which two shapes to test properly with the tape.

Face shape is a proxy for where light lands and where volume reads. That makes it a decision input for three practical things, and nothing else.

• Haircut volume. The question is only ever where to add height and where to keep width. A long face wants width at the sides and no height on top; a wide face wants the opposite. Whatever you pick, the upkeep is heat, and the American Academy of Dermatology is blunt about that: limit blow drying and irons, and use low or medium heat with a protectant.
• Glasses width. Frames read as a horizontal line across the widest third of your face, so their width relative to your cheekbones matters far more than their colour.
• Camera angle. The direction you turn and the height of the lens change your apparent jaw width more than any edit does.

That is the whole return on the exercise. It is a cosmetic classification, not a health assessment, and the medical libraries that cover hair and skin, Cleveland Clinic among them, do not carry face-shape entries at all. If you want to act on the answer, the next step is testing a cut against it in the hairstyle changer rather than committing in a salon chair.

The measurement is worth two minutes because it converts an argument about adjectives into a decision about width.

— Maya Chen

Optical shops sell this as a personality quiz. It is closer to carpentry: pick the frame geometry that contradicts your dominant line, and keep the frame no wider than your cheekbones.

One rule beats the table when they disagree: contrast the jawline. A soft jaw takes a hard frame, a hard jaw takes a soft one. Everything in the table is a specific case of that. The fashion desks at Allure and Cosmopolitan will give you the seasonal version of this advice; the geometry underneath it does not change with the season.

The conventional answer is the oval face shape, and the reason given is circular: it is treated as the default because most styling advice is written to move other shapes toward it. That is a statement about the advice, not about faces.

There is a practical grain of truth in it. A face whose forehead, cheek and jaw widths are close together, at about one and a half times its own width in length, does not fight most haircuts or most frames, so the decision moves to hair texture and upkeep instead. We wrote the measurement detail up separately in the oval face shape post.

What we will not do is rank faces. No published figure ranks facial geometry by attractiveness, and the ones that circulate online are unattributed. Grooming and beauty desks such as GQ and Self have largely dropped the ranking framing for the same reason. Use the shape to choose a haircut, then stop.

Run the same face through two tools and you will often get two answers. Three things cause it, and all three are in your control.

• Lighting. A single overhead light throws shadow under the cheekbones and narrows the jaw, which pushes a square reading toward heart.
• Angle. A phone held below eye level shortens the forehead and widens the jaw, moving oval toward square or round.
• Hairline coverage. A fringe hides the top measurement entirely, and the tool has to guess where your forehead ends.

There is a fourth, less obvious one, and our own run showed it. Both faces we tested came back as oval, and the reference strip on the result card carries five silhouettes, oval, round, square, heart and diamond, with no oblong option. A classifier that cannot output your shape will hand you the nearest one it has. The same caution applies to the general-interest health explainers on sites like Healthline: they describe the categories, they do not measure your face. That is worth knowing before you take any face shape app verdict as final, ours included.

So the honest workflow is a hybrid. Measure once by hand and keep the four numbers. Use the app when you want the picture, the reference lines and something to send to a stylist. If you searched what is my face shape app expecting a lab result, the tape is the lab; the app is the diagram. Both were fast in our test: in LazyKiwi workbench data (August 2026), the median time from signing up to a first generation is about a minute and a half.

If you want to run a batch of angles rather than one photo, the same tools sit inside the image workbench, where you can queue several runs and compare the cards side by side in My Image.