Skip to content
The Face Shape Detector
Color and style

Skin Tone Analyzer

Sample your forehead and cheek color from a photo to estimate a Fitzpatrick type and a warm, cool, neutral or olive undertone.

Loading tool…
  • 468

    Landmarks mapped

  • 0

    Data collected

  • Free

    Always, no account

  • Any

    Device or browser

3 simple steps

How the skin tone analyzer

What happens from the moment you start, in order.

  1. 1

    Sample three skin patches

    MediaPipe locates the center of your forehead and both cheeks below the outer eye corners, then each patch is averaged with highlights, shadow edges and blemishes discarded.

  2. 2

    Convert to perceptual color

    The combined RGB average converts into CIE L*a*b* so lightness separates from the two chromatic axes, which is what makes depth and undertone independently readable.

  3. 3

    Estimate type and undertone

    Lightness maps onto the Individual Typology Angle for a Fitzpatrick I to VI estimate, while the warm-to-cool axis and the green component name your undertone family.

What you get

Built to be checked

Every reading comes with the measurement behind it.

Undertone including olive

Olive skin is detected through the green component instead of being squeezed onto a warm-to-cool axis, where most foundation ranges mismatch it as too pink or too orange.

Color sampled on device

Pixel sampling and Google MediaPipe landmark detection both run in your browser. No photo is uploaded, stored, or transmitted anywhere at any point.

Foundation and season match

Your result maps to the W, C and N shade codes brands print on the bottle, plus the color-season families your undertone and depth point toward.

Clear about its limits

This is a cosmetics shade aid, not a dermatological assessment. It reads average pixel color only and cannot evaluate skin health in any way.

A skin tone analyzer samples the average color of facial skin from a photograph and converts it into two cosmetic descriptions: an estimated Fitzpatrick type and an undertone family. Both are shade-matching aids. They exist so you can pick a foundation, a lipstick family or a wardrobe palette with better odds than guessing.

Three regions supply the sample: the center of the forehead and both cheeks below the outer eye corner. Each patch is averaged with outliers discarded, so a stray highlight or a blemish cannot drag the figure. The combined average converts from image RGB into the CIE L*a*b* perceptual space, where lightness separates cleanly from the two chromatic axes. Lightness maps onto the Individual Typology Angle, a colorimetric measure standard in cosmetic science, which yields a Fitzpatrick I to VI estimate. The warm-to-cool axis and the green component yield one of four undertone families: warm, cool, neutral or olive.

Say the limit plainly. This reads average pixel color and cannot be a dermatological assessment. It does not examine texture, lesions, moles or pigmentation patterns, and it cannot identify or rule out any skin condition. Concerns about a mark, a change in pigmentation or how your skin responds to sun belong with a dermatologist or physician.

Google MediaPipe Face Mesh locates the three patches through its 468-landmark topology, and every step runs inside your browser. No photograph is uploaded, received or stored.

Ahead: why the Fitzpatrick scale needs context, its six rows set against their original purpose, the four undertone families and the checks that agree with them, how to turn a result into foundation and wardrobe choices, the lighting rules for a color-accurate frame, and why white balance caps the exercise.

Why the Fitzpatrick scale needs context

The scale was designed in 1975 for one narrow job: predicting how skin responds to ultraviolet exposure during phototherapy dosing. It was built around burning and tanning rather than appearance, and it carries documented limitations for darker skin. Cosmetics adopted it because it was the shorthand available.

That history changes how a type should be read. A photograph cannot observe how your skin reacts to sun. It measures lightness and infers a position from it, so the estimate reconstructs the scale's original variable instead of watching it. Your own experience of burning and tanning is better evidence, and it should override anything here. The rows below are a range to place yourself on, not a fact about your biology.

The six types and their original purpose

TypeAppearanceUV response
IPalest, often freckledAlways burns, never tans
IIFairBurns easily, tans minimally
IIILight to mediumSometimes burns, tans gradually
IVOlive to moderate brownRarely burns, tans readily
VBrownSeldom burns, tans deeply
VIDeeply pigmented brown to blackAlmost never burns

The right-hand column is what the scale exists to describe, and no camera reaches it. Lightness puts you near a row. It does not confirm the row. Depth is also half of a shade match, since two people at one depth can need different foundations, and the reason is undertone.

The four undertone families

Undertone is the cast beneath the surface, and it holds steady even when surface depth shifts with sun exposure. It decides whether a correctly matched foundation still looks slightly wrong.

  • Warm skin carries a golden, peach or yellow cast. Gold jewelry tends to look more natural than silver, and cream-based neutrals beat stark white.
  • Cool skin carries a pink, red or blue cast. Silver and platinum sit well against it, and jewel tones read clean rather than harsh.
  • Neutral skin carries no dominant cast. Both metals work, and the gain is a wider band of usable shades.
  • Olive skin carries a distinct green or gray cast, at any depth from light to deep. It gets mismatched often, because most foundation ranges run a warm-to-cool axis with no green correction, so shades land too pink or too orange.

Two traditional checks agree with these families reasonably well. Look at the veins on your inner wrist in daylight: greenish points warm, blue or purple points cool, a mix points neutral. Then hold pure white paper beside your bare face. Warm skin looks yellow against it, cool skin looks pink, olive skin looks faintly gray. Both checks feed the same practical decisions.

Turning a result into foundation and wardrobe choices

Foundation matching works in two steps. Match depth to the type estimate, then match undertone to the letter code most brands print on the bottle: W for warm, C for cool, N for neutral. Test the candidate along the jawline in daylight, not on the back of your hand, which usually differs in depth from your face.

Seasonal color analysis takes undertone as one of three inputs, alongside value (how light or dark your overall coloring runs) and chroma (how saturated it is). Warm undertone points toward the spring and autumn families, cool toward summer and winter. Olive most often lands in the deep or soft groupings.

For clothing, the shades nearest your face carry the most weight. Cool undertone is flattered by jewel tones and true whites, warm undertone by rust, olive, camel and cream. Every call depends on the photograph reporting your color honestly.

Lighting rules for a color-accurate frame

Color measurement is far more sensitive to capture conditions than any geometric one, because the camera decides what counts as white before you see the image.

Use indirect daylight. Stand near a window, facing it, when the light isn't strongly orange. Switch every indoor bulb off, because mixing daylight with warm artificial light hands the camera two white points to split between.

Turn off beauty and smoothing filters, plus auto-enhance if your camera applies it. Those features adjust saturation and warmth by design. Photograph bare skin: no foundation, tinted moisturizer, bronzer or SPF that leaves a cast. Keep saturated clothing and painted walls away from your face, since both bounce their own cast onto your skin. One variable still stays outside your control.

Why white balance caps the whole exercise

White balance is the largest single error source, even under good conditions. A phone camera infers the color of the light from the scene, and it can land several hundred kelvin off, which shifts the entire image warm or cool and moves the undertone reading directly. That is why the same person can get two different results from two frames taken minutes apart in different rooms.

Other variables stack on top. Recent sun exposure changes surface depth without touching undertone, so a tanned frame may place you a type above baseline. Redness from heat, exercise, flushing or irritation pulls the chromatic average toward cool. Screen calibration changes how you perceive a result, not the measurement.

Treat the output as a starting point. It narrows a search from thirty foundation shades to roughly three. Confirm the final pick in person, in daylight, on your own skin. No photo tool can take that last step for you.

Two descriptions, three patches, one browser tab. The estimate comes from average pixel color converted into CIE L*a*b*, with lightness mapped through the Individual Typology Angle onto a Fitzpatrick type and the chromatic axes mapped onto a warm, cool, neutral or olive family. The scale needs context because it was built for phototherapy dosing rather than appearance, the type table reports a UV response no camera can see, and undertone decides what depth alone cannot. Foundation codes, seasonal families and wardrobe choices follow, indirect daylight protects the reading, and white balance still caps accuracy. Nothing here is dermatological. Open the skin tone analyzer in daylight and treat the result as a shortlist.

Questions

Skin Tone Analyzer FAQ

The questions people ask most often about this tool.

How do I find my skin undertone?

Two checks settle most cases. Look at the veins on your inner wrist in daylight: greenish points warm, blue or purple points cool, a mix points neutral. Then hold pure white paper beside your bare face. Warm skin looks yellow against it, cool skin looks pink, and olive skin looks faintly gray, which is a fourth family in its own right.

What is the Fitzpatrick scale?

The Fitzpatrick scale is a six-point classification created in 1975 to predict how skin responds to ultraviolet light during phototherapy dosing. Type I always burns and never tans, while type VI almost never burns. It was built around sunburn behavior rather than appearance, and it has documented limitations for darker skin, so any estimate is an approximate position on a range.

Is my photo uploaded to check my skin tone?

No. Both the landmark detection and the pixel sampling happen inside your browser using Google MediaPipe, so the image is processed on your own device. Nothing is sent to a server, nothing is stored, and no copy is retained. Closing the page clears the photo from memory and leaves nothing behind for anyone to access.

Can this tool tell me about my skin health?

No, and it should not be used that way. It reads average pixel color from three small regions and nothing else. It does not examine texture, moles, lesions or pigmentation patterns, and it cannot identify or rule out any skin condition. Any concern about a mark, a change in pigmentation or sun sensitivity belongs with a dermatologist or physician.

Which affects the reading more, lighting or camera quality?

Lighting wins, and by a wide margin. Automatic white balance is the largest single error source: the camera guesses the color of the light and can land several hundred kelvin off, shifting the whole frame warm or cool and moving the undertone directly. Sensor quality matters far less. Recent tanning, flushing and saturated clothing near your face also skew results.

How should I light the photo for a color-accurate result?

Use indirect daylight. Face a window and switch off every indoor bulb, because mixing daylight with warm artificial light hands the camera two white points to split between. Turn off beauty filters and auto-enhance, remove foundation, bronzer and tinted SPF, and keep saturated clothing or painted walls away from your face.

Start with your face shape

Most styling decisions follow from the shape of your face. Run the detector, then work outward from the result.