Telling warm from cool, and the four seasonal color types

1. What personal color is

Personal color is about whether a given color, held near the face, makes that person's skin read more even and clear. No color is good or bad in itself; what is being judged is the pairing with a particular complexion. That is why the same red gives one person life and makes another's face read separately from their clothes.

The roots are in color teaching. Mapping colors onto the feeling of four seasons is usually traced to Johannes Itten, who taught color at the Bauhaus, and carrying that over into clothes and makeup for individual people was spread widely by Carole Jackson's 1980 book Color Me Beautiful. After that book sold, the four seasons became the shared language of the cosmetics counter.

Because it began in books and classes rather than in medicine or measurement, this field has no agreed standard. Studios use different numbers of categories, anywhere from four to sixteen, and the same person really does get different verdicts in different places. Knowing that makes the limits of a self-test easier to accept.

The practice became notably granular in South Korea, where the same classification grew into an industry of paid analysis appointments. Finer subdivisions splitting the four into eight, twelve or sixteen, a settled order for the draping cloths, and a path from the verdict through to specific cosmetics all hardened into one procedure. That is why Korean-language material on this subject is often more specific than English-language material covering the same ground.

2. The basic rules: two axes, four types

The first axis is temperature. A yellow cast under the skin is called warm, a blue cast cool. This is the undertone, and it is a different axis from how light or dark the skin looks on the surface. There are warm people with very fair skin and cool people with deep skin.

The second axis is depth: whether your natural hair and eye color, and the contrast within your face, sit at the light end or the deep end. Crossing the two axes gives four boxes. Warm and light is spring, warm and deep is autumn, cool and light is summer, cool and deep is winter.

Each season name carries a color character with it. Spring is clear and bright, summer soft and pale, autumn muted and deep, winter clear and deep. So within cool, summer carries a pink softened by a step while winter carries a vivid magenta.

3. The rules this page counts by

There are eight questions in two groups: five on undertone (vein color, how skin reacts to sun, gold against silver, pure white against ivory, which lipstick family has felt right) and three on depth (natural hair color, eye color, contrast within the face). Both groups are odd-numbered to reduce the chance of a tie.

The verdict is nothing but counting. More warm than cool among the undertone questions reads warm; more deep than light among the depth questions reads deep; crossing the two points to one of the four boxes. "Not sure" answers are not discarded, they are counted separately and reported on screen. There are no weights and no corrections, and the same answers always give the same result.

If either axis ties, no season is named. Rather than forcing a pick, the remaining two candidates (or all four) are shown as they are. And when the undertone gap is one or less, or depth is tied, the page says "your answers are split" before anything else. A weak lean should not be dressed up as certainty.

The only numbers offered are counts of your answers. No accuracy or probability figure is produced: there is nothing behind such a figure, and attaching one would make the words "for reference" meaningless. The hex codes in a palette are likewise values picked to show a family on screen, not an official chart.

There is also a route that measures a photo. Supplying a photo of your face and tapping the middle of a cheek gathers the pixels at that spot into an average color, which is converted to CIELAB coordinates and measured on two values: ITA degrees for lightness, and the hue angle separating the red side from the yellow side. Pixels that are too dark, specular highlights, and near-grey pixels are discarded, and the remaining count is shown on screen. If the sample is too small or the color too weak, the page declines to judge rather than produce a number.

This page does not detect your face. With neither a face-detection model nor a skin-classification model, where the skin is has to be pointed out by a person. Rather than talk about an artificial intelligence we do not have, we ask once. Of the two boundaries, the 28 degrees for ITA is taken straight from the bands dermatology literature uses, while the 50 degrees for hue angle is a value we chose. That is why the measured values, the hex code, the ITA figure, the hue angle and the usable pixel count, are shown alongside the verdict rather than only the verdict.

Neither the answers you pick nor the photo you supply reaches the server. The photo is drawn onto a canvas inside the browser and only its pixels are read; there is no upload request at all. Nothing is stored in this browser either, so a refresh clears it. The server learns only three facts: that the page was opened, that a result appeared once, and that a link was copied.

4. Practical points for judging at home

Light changes the answer. Fluorescent light adds a blue cast and incandescent light a yellow one, so the same face reads as a different tone. Daylight by a window on an overcast day is the most even. Take makeup off, and push your hair back off your face.

Holding cloth up often works better than answering questions. Alternating a pure white towel and an ivory towel under your chin in front of a mirror usually shows one side reducing the shadows on your face and the other making it read sallow. Not being able to see a difference is also an answer; plenty of people really are on a border.

If judging alone is hard, ask a person rather than a photo. Cameras and processing shift color, which clouds the judgement rather than helping it. Putting on a gold chain and then a silver one and asking whoever is nearby which looks better can be more accurate than your own eye.

5. A little deeper

There are systems that split the four further. Some studios divide even within spring, separating the light end from the clear end, out to eight, twelve or sixteen categories. This page stops at four. Splitting sixteen ways off eight questions would leave each box resting on too little, which amounts to manufacturing a precision that is not there.

You will also hear muted and clear. That is chroma, whether a color is muddied or vivid. This page does not ask about chroma separately, because we did not find a reliable way to judge it through self-test questions. Conventionally spring and winter sit at the clear end and summer and autumn at the softer or muted end, so the depth axis tends to move in much the same direction.

Professional draping works through dozens of cloths held under the face in sequence, which allows comparison between pieces that differ only slightly within one family, and so reveals more than a questionnaire. It is worth knowing as well that the field has no accredited qualification and no agreed measure of accuracy. We did not find peer-reviewed research comparing self-tests with professional analysis.

6. Where people get stuck

Answers splitting roughly in half is the most common case. When that happens this page names no season and shows the candidates instead. Anyone on a border does best starting from the colors the two palettes share: between spring and summer, for instance, a mid-brightness pinkish brown straddles both.

Some people cannot see their veins at all. Thicker skin does not let them show through, and then "not sure" is the right answer on that question. Because "not sure" is counted rather than discarded, you can look back later at which question you could not call.

If your hair is dyed, look at the roots. If that is hard to find, the hair color in childhood photographs is a useful reference. Take contact lenses out for the eye question, and while you are there notice how sharply the iris stands against the white, which feeds the contrast question.

Your result may differ from an analysis you had before. Studios use different criteria, and this page looks at eight questions only. We have no grounds for declaring which is right, so if you have actually worn and tried things, trust that experience over this.