Oblong face shape

How this site represents an oblong face

The classifier holds one set of reference ratios per shape and picks whichever is closest to yours across all four at once. These are comparison points, not averages of people with that face shape, and not a definition anyone else is obliged to use.

For oblong the reference values are: length over width 1.62, jaw over cheekbone 0.80, forehead over cheekbone 0.88, and lower jaw over cheekbone 0.680.

Set those beside the oval reference values and the structure of the comparison shows. The jaw ratio is the same in both at 0.80, and so is the forehead ratio at 0.88. What differs is length — 1.62 against oval's 1.45 — and, by much less, the lower jaw ratio, 0.680 against 0.664. Length carries most of the separation, and the lower jaw ratio still participates: with the other ratios held fixed, moving it alone across that small range is enough to change which of the two labels a borderline face receives.

The oblong reference values, and how the label is assigned

What scale these numbers are on

This matters before you reach for a ruler. The figures above are not raw divisions, and they are not on one shared scale. The first three mesh ratios use separate scale factors: length over width × (1.45 / 1.152), jaw over cheekbone × (0.80 / 0.775), and forehead over cheekbone × (0.88 / 0.670). These map the canonical mesh reference values used in the code to 1.45, 0.80 and 0.88 respectively. The lower jaw ratio is unscaled.

The value 1.152 is the canonical mesh reference ratio used in the code, not a promised ruler measurement. Measurements taken by hand with different endpoints cannot be compared directly with these classifier values. So use these figures to understand which reference set you are nearest, not as targets to reproduce with a ruler. The hand method, with its own caveats, is set out in how to tell your face shape.

Where the line between oblong and oval falls, and what happens when you land near it, is worked through on oblong vs oval face shape.

Oblong, long face, rectangle

These terms overlap, and usage varies between sources. Here, oblong is the name of a classifier category defined by the ratios above. Long face is a broader description of appearance rather than a measured category. Rectangle appears in some styling systems, where it can carry the extra sense of a long face with a more angular outline.

Worth checking which sense a page means before following it, since the two can point in different directions. This site has no rectangle category: a long face with a wide, angular jaw is compared against the square reference set as well as the oblong one, and whichever is nearer across all four ratios becomes the label.

What the classifier does and does not look at

It compares four width and length ratios. It does not measure jaw angle, the curve of the jawline, chin shape, or how far the cheekbones project. Those features are on your face; they are not what these ratios encode. A description that turns on jaw angle rather than jaw width is describing something the numbers here do not capture.

It also compares against all six reference sets at once, not against oval alone. Each of the others is a different combination rather than a single trait: heart pairs a jaw ratio of 0.76 with a forehead ratio of 0.96, diamond pairs the same 0.76 jaw with a much narrower 0.80 forehead, and round and square differ from each other mainly in jaw width at a shorter length. So a single word like "narrow jaw" does not identify a shape on its own.

What moves the measurement without changing your face

Every ratio here comes from a two-dimensional photograph, so the numbers describe the image as much as the face in it. Three things are worth controlling before treating a result as settled.

• Head position. Turning or tilting your head changes the projected distances between landmarks, and the ratios calculated from them. The scanner screens for this by comparing how far your nose sits between your cheekbones and where your eyes sit down the length of your face, and declines photos that fall outside its range. Before measuring it levels the eye line, which compensates for in-plane head roll; the yaw and pitch checks screen photos rather than correct them, so an accepted photo still carries those.
• Camera distance. Distance from the lens changes apparent proportions through perspective. If you are comparing two photos, keep the distance similar.
• What is visible at the edges of the face. Face length here is the vertical separation between forehead mesh point 10 and chin point 152 after the eye line is levelled; point 10 is not the hairline. Cheekbone width uses points 234 and 454, jaw width points 172 and 397 near the jaw angles, and lower jaw width points 136 and 365 between the jaw angles and the chin. Anything that obscures those regions — hair across the forehead, a beard changing the visible outline of the lower face — makes the landmarks harder to place. What changes is where the endpoint is estimated, not the face underneath it.

These are distances between estimated mesh landmarks, with separate scale factors applied to the first three ratios; they are not direct measurements of the bone beneath. Other measurement traditions use different endpoints — some facial measurement systems work from the nasion — so a length taken elsewhere is not directly comparable to this one.

What to do with the label

A shape label is a description, not a problem statement. Nothing about an oblong face needs correcting, and the classifier makes no claim about health or attractiveness. If you want to explore ways of making your face appear shorter, the styling guides below offer ideas. That goal is a personal preference; these ratios do not predict the effect of a particular cut or frame.

• Specific cuts, for women and men: oblong face shape hairstyles
• Frame ideas: glasses for an oblong face
• If you measured close to oval: oblong vs oval face shape
• Measuring the ratios by hand from a photo: how to tell your face shape

Or run the face shape tool and read your own ratios rather than estimating them in a mirror.