Nose Shape Detector: What Nose Do I Have, and Is It Big?

The test finds 478 points on your face and takes two measurements of your nose: its width, between the outer edges of your nostril wings, and its length, from the bridge between your eyes to the base of your nose. Both are measured in units of the gap between your inner eye corners, the face’s own ruler in the rule of fifths, and the size is the width times the length. The points are read on the face model’s 3D mesh, on your photo and on its mirror image, averaged, because the model does not place a face and its mirror image the same (the photo alone and its mirror alone put the size 7 points apart at the median).

Each reading is compared with 165 front-facing photos of public figures, 84 men and 81 women, measured exactly the same way: press, event and sports photos taken by photographers, with the head straight and the lips together, because an open smile pulls the nostril wings about 6.5% wider. A reading becomes "wider than 64% of women’s noses" instead of a raw ratio. Men and women are compared separately because men’s noses read wider for their eye gap; Everyone compares you with all 165. Measuring against the eye gap has one catch: wide-set eyes make a nose read smaller, close-set eyes larger.

The size bands split the reference faces of each sex into about 15% at each end, 20% next to them and 30% in the middle, so an average result is the most common one.

For width and length, the middle half of men reads 1.22 to 1.33 and 1.39 to 1.55 eye gaps; the middle half of women reads 1.13 to 1.24 and 1.36 to 1.49. Above the middle half reads wide or long, below it narrow or short.

A camera close to your face enlarges whatever is nearest to it, and the nose is the nearest part. Ward and colleagues (JAMA Facial Plastic Surgery, 2018) worked out that a photo from about 30 cm makes the base of the nose look about 30% wider than one from 1.5 m. The reference faces were photographed from a distance, so a selfie compared with them would read big for the wrong reason.

The test handles it in two ways. It measures the nose against the gap between your eyes, which sits only a little behind the nose, instead of the width of your face, which sits far behind it: simulated on the reference faces’ own 3D landmarks, a photo from 50 cm read the size 46 points high against the width and height of the face, and 11 points high against the eye gap. Then it works out how far away the camera was and moves every point back by its own depth, read from the face model. The distance comes from your iris, which is close to 11.7 mm across in adults (11.71 mm on average, give or take 0.42 mm, in 390 adults: Rüfer and colleagues, Cornea, 2005): how many pixels it spans says how far away the camera was, once the camera’s focal length is known.

The focal length comes from one of three places. A photo from this page’s camera uses a typical front camera: the middle of four with published fields of view (a Logitech C920 webcam, a Galaxy S24, an iPhone 13 and a Pixel 8), which reads the distance from 8% too near to 20% too far depending on the camera. An uploaded photo uses the focal length its camera wrote into the file, when the file still has it and has not been cropped. Without either, the test assumes an arm’s length selfie, 45 cm, the arm’s length of Pressler and colleagues (Plastic and Reconstructive Surgery, 2022), and the result says it assumed one.

In the same simulation, with the iris varied by that spread: a selfie from this page’s camera reads the size within 3.1 points of a far-away photo in 9 of 10 (0.7 at the median), an uploaded photo that keeps its focal length within 0.9, and the arm’s length assumption within 7.9 for selfies, but about 8 points small at the median for a photo taken from 1 or 2 m. On real photos, Pressler and colleagues measured noses 6.4% longer at 30 cm and 4.3% longer at 45 cm than at 1.5 m; the same geometry reads 4.9% and 2.8%, the same direction and a little less, so a close photo may still read a little long. For the truest reading, have someone take the photo from 1 m or more, or prop your phone and use the timer.

The named types split into two families. Profile types (Greek, Roman, aquiline, snub or button, celestial) are about the bridge and the tip seen from the side; front types (narrow, wide or Nubian, bulbous) are about width and the tip seen from the front. The chart draws each one with the straight line the side reading measures against, and says what this test reads for it.

A front photo cannot see a bridge, so a profile type needs a side photo. Without one, the result names your front shape (for example "Narrow and long") and your size, and leaves the bridge out instead of guessing it.

On a side photo, the test traces the outline of your face with the same on-device model as the side profile analysis, finds the deepest point between your eyes and your nose tip, and draws a straight line between them. It then measures how far your bridge stands in front of that line, or sits behind it, between 20% and 50% of the way down, where the bridge is and the tip has not started. A bow of 5.0% of the line’s length or more reads Roman, a scoop of 1.0% or more reads Snub, anything between reads Straight (Greek).

There is no published set of side photos labeled by nose type to check this against, so the cut-offs were set on 24 generated side photos, one prompt per type, each labeled by eye before any reading was taken. The outline was traced on 22 of them; of the 19 with a sure label, 17 read as labeled, and the two that did not were snub noses with a slight scoop that read Straight. Across 8 versions of each labeled photo (rotated, cropped, half size, compressed), 141 of 152 were traced and 124 read as labeled. Generated studio photos are cleaner than most real ones, so take the side photo against a plain wall.

The tip is not read. The turn of the tip is what separates an aquiline nose from a Roman one and a snub nose from a celestial one, and on the test photos it did not tell them apart, so the result names the bridge and says the tip is not measured.

The rule of fifths, an art and surgery canon, divides the face across into five eye widths and puts the nose in the middle one: as wide as the gap between the inner corners of the eyes. Measured on photos, every one of the 165 reference faces read wider than its eye gap, the narrowest at 1.04 times, because the points sit on the groove at the outer edge of each nostril wing and at the visible inner corner of each eye, not where a caliper would go. So the test never grades you against the rule’s 1.0; it compares your ratio with faces read the same way.

The nose also sets the middle of the facial thirds. The facial thirds test measures how your middle third compares with the other two, the philtrum ratio splits the third below your nose, and the golden ratio face calculator checks the classical ratios, including mouth width over nose width, which the nose to lip ratio guide explains.

You need one front photo; the side photo is optional and adds the bridge. Both are read in your browser and never uploaded.

A tipped head changes how long a nose reads, by about 3.3% per 10° on the reference faces; the test reads the tip from your face and takes it out. Take two photos the same way; if they land in different bands, the photo is the variable, not your nose.

The shape is set by bone and cartilage, which exercises and nose-shaping clips do not change in adults. What you can change is how it reads. Contour makeup down the sides of the bridge is used to make it read narrower, and a highlight that stops short of the tip to make it read shorter; stylists balance a prominent nose with volume at the sides of the hair, a fuller beard or glasses with a heavier bridge. And a photo from further away shows it closer to its real size.

Filler injected by a qualified practitioner can smooth a scooped bridge or disguise a small hump, which is why it is sold as a non-surgical nose job, but it adds volume and cannot make a nose smaller. Making a nose smaller, narrower or straighter is rhinoplasty, and whether it suits your face is a decision for a board-certified facial plastic surgeon after an examination, not for a percentile.