Real proportional measurements
Read the numbers behind the label: length-to-width ratio, three facial widths and your gonial angle in degrees, rather than a category name on its own.
Measure forehead, cheekbone and jaw width against face length to name which of the seven face shapes your outline matches.
468
Landmarks mapped
0
Data collected
Free
Always, no account
Any
Device or browser
3 simple steps
What happens from the moment you start, in order.
Face the lens at eye level, hair pulled back, expression neutral. Your image stays in the browser while 468 MediaPipe landmarks map across hairline, temples, cheekbones, jaw corners and chin.
Forehead width is read at the brow ridge, cheekbone width across the zygomatic arches, jaw width between the gonial angles, and total length from hairline midpoint to chin base.
Your length-to-width ratio and the taper between the three widths narrow the field, then the gonial angle separates square from round. Styling guidance arrives with the result.
What you get
Every reading comes with the measurement behind it.
Read the numbers behind the label: length-to-width ratio, three facial widths and your gonial angle in degrees, rather than a category name on its own.
Detection runs inside your browser through Google MediaPipe. No image reaches a server, gets stored anywhere, or is shared with anyone at any stage.
Oval, round, square, oblong, heart, diamond and triangle each carry explicit numeric thresholds, so borderline outlines get flagged instead of forced into a box.
Every result carries haircut, fringe, eyewear and neckline guidance built on the counterbalance principle stylists and opticians have worked from for decades.
Face shape classification (facial morphology typing) sorts the outline of a human head into one of seven named categories using four measurements. The seven are oval, round, square, heart, diamond, oblong and triangle. The scheme grew out of portrait painting, then out of twentieth-century hairdressing and optical dispensing manuals, where stylists needed fast vocabulary for balancing a client's proportions.
Four numbers do the work. Three are horizontal chords, taken across the brow ridge, the zygomatic arches and the gonial angles. The fourth is a vertical drop from hairline to chin base. Divide that drop by the widest chord and you have the length-to-width ratio: readings near 1.0 are short and compact, readings past 1.5 are elongated. A fifth value, the interior angle at the gonial corner, settles the cases the ratio leaves tied.
Google MediaPipe Face Mesh supplies the geometry. Its 468-landmark topology fixes sub-pixel positions for the hairline, temples, cheekbones, jaw corners and chin, and the whole computation happens inside your browser tab. No image is uploaded, received or stored.
Below you'll find what the seven categories are for, the measurements and taper profile built on them, a side-by-side comparison of every shape, how stylists convert a label into haircut, fringe, eyewear and neckline choices, the photo conditions that shift a reading, and the limits of measuring a three-dimensional skull from a flat picture.
Categories are a starting point for styling, not a description of who you are. Two people given the same label can look nothing alike, because the label compresses four numbers and discards skin, expression, coloring, feature size and profile depth.
They also sit on a continuum, and plenty of faces land within a few hundredths of a boundary. What makes the reading repeatable is the set of measurements underneath it.
Measurements come as one vertical and three horizontal, each taken from named skeletal landmarks rather than eyeballed.
Those four also produce the taper profile, which compares the three widths and reports whether the outline narrows upward, narrows downward, or runs parallel. Taper plus the ratio resolves five categories on its own.
The remaining pair needs the gonial angle. Below roughly 120 degrees the jaw reads square and defined. Above roughly 135 degrees it reads soft and rounded. That angle is the only thing separating a square outline from a round one when their proportions are near identical, which is why both sit together in the comparison below.
| Shape | Length to width | Width relationship | Jawline signature |
|---|---|---|---|
| Oval | 1.45 to 1.6 | Cheekbones widest, forehead a little broader than jaw | Gently curved, no hard corner |
| Round | 0.95 to 1.15 | All three widths close to equal | Soft, obtuse gonial corner |
| Square | 0.95 to 1.2 | All three widths close to equal | Sharp, near right-angle corner |
| Oblong (rectangle) | Above 1.5 | Widths near equal, outline long | Squared, softened by the extra height |
| Heart | 1.3 to 1.5 | Forehead clearly widest, jaw narrowest | Narrow, pointed chin |
| Diamond | 1.4 to 1.6 | Cheekbones dominant, forehead and jaw both narrow | Angular, tapered to a point |
| Triangle (pear) | 1.2 to 1.45 | Jaw widest, forehead narrowest | Broad and heavy at the base |
Oval is the reference the other six get described against: moderately long, widest at the cheek, curving gradually from temple to chin. Oblong is a square stretched vertically. Heart and triangle mirror each other, one tapering down and one tapering up. Diamond turns up least often, since it needs both the forehead and the mandible narrower than a pronounced cheekbone line. Each label drags styling conventions along with it.
Styling conventions run on one principle: counterbalance. Add apparent width where the outline is narrow, add height where it's short, soften where the line is hard.
None of it survives a bad photograph, which is where most disputed results begin.
Photographs decide the numbers, so five capture conditions matter more than the algorithm does, listed here in descending order of impact.
Keep your head level and square to the lens. Ten degrees of yaw compresses one side and shifts every horizontal reading, and projection sets the ceiling on accuracy.
Accuracy is capped by the medium. A photograph is a flat projection of a three-dimensional skull, so depth gets inferred rather than measured. How far a cheekbone comes forward, how deep a chin sits: neither is in the picture, and two people with different profiles can share one frontal outline.
Borderline readings are common and are not errors. A ratio of 1.47 sits on the oval-to-heart boundary, and a small change in lighting or pose tips it either way. Two labels from two photos means you carry features of both.
Reported reliably: your proportions, which region is broadest, which is narrowest, and how the thresholds relate to those figures. Not reported at all: health, ancestry, character or attractiveness. The numbers are objective. The aesthetic conclusions drawn from them are conventions, and any convention you dislike can be ignored.
Seven categories, four measurements, one browser tab. The system compresses a head outline into three horizontal chords, one vertical drop, a taper profile and a gonial angle, then names the closest of oval, round, square, heart, diamond, oblong or triangle. The table shows how each differs from its neighbors, the styling section converts a label into haircut, fringe and eyewear choices, and the capture checklist protects the numbers from hair, expression, camera height and lens distortion. The limits stay on the page: flat projection, soft boundaries, nothing medical. Upload a level frontal shot to the face shape detector and read the ratios, not just the name at the top.
Questions
The questions people ask most often about this tool.
Seven: oval, round, square, oblong, heart, diamond and triangle. Oval sits widest at the cheek with a gradual curve to the chin. Round and square share their proportions and split on jaw angle. Oblong is a square stretched vertically. Heart and triangle mirror each other, one tapering down and one tapering up, while diamond needs a narrow forehead and jaw under dominant cheekbones.
The length-to-width ratio divides total face length, hairline midpoint to chin base, by the widest of the three horizontal measurements. Readings near 1.0 describe a short compact outline. Readings past 1.5 describe an elongated one. It is the first value any classification system computes, and it narrows seven categories down to two or three before any other measurement is applied.
No. The face mesh is computed locally by Google MediaPipe running inside your browser tab. Your image never travels to our servers, is never written to a database, and clears from memory when you close or reload the page. Nothing is retained after your session ends, and no account is needed to use the tool.
Yes, gradually. Facial fat distribution shifts with age and body weight, which mainly moves apparent cheek and jaw width. Bone structure changes far more slowly, with some remodeling of the gonial angle across decades. A result that changes within weeks is almost always caused by weight, swelling or photo conditions rather than by skeletal change.
A photo wins on repeatability. Landmark detection places the same anatomical points every time, while a tape slips against soft tissue and depends on how hard you press. A tape wins on depth, since it can follow a curve that a flat image cannot see. For proportional ratios, which is what classification needs, the photo method is the more consistent of the two.
Four causes account for almost every disagreement. Hair crossing the temples narrows the detected forehead. A smile widens the cheekbone line. Holding the phone below eye level enlarges the jaw. And a wide-angle front lens exaggerates whatever sits nearest the sensor. Retake the shot at eye level, an arm's length away, hair back, under even frontal light.
Keep exploring
Mirror your facial landmarks about a computed midline and score how closely the left and right halves correspond.
Try it freeCompare your facial proportions against phi, roughly 1.618, across seven classical ratios and read the full per-ratio breakdown.
Try it freeMeasure canthal tilt, lid crease and eye spacing from a photo to place you among the ten eye shape descriptors.
Try it freeMost styling decisions follow from the shape of your face. Run the detector, then work outward from the result.