How to Do Color Analysis at Home With Just Your Phone Camera

You do not need a consultant or a kit to get a usable read on your color season. You need daylight, a phone, a white top and about ten minutes. What you do need is a method, because doing this badly is easy and produces a confident wrong answer.

Here is the whole process.

Before you start: what you are actually looking for

Color analysis sounds mystical and is not. It comes down to three measurable properties.

Depth. How light or deep your overall coloring is, taken together, skin, hair, eyes. A platinum blonde with pale skin sits at one end; deep brown hair with deep skin sits at the other.

Undertone. Whether your skin leans warm (more yellow and golden) or cool (more pink and blue), or sits neutral between them. This is not the same as how light or dark you are. Deep skin can be cool and fair skin can be warm.

Contrast. The size of the gap between your features, mostly hair against skin. Black hair with fair skin is high contrast. Light brown hair with medium skin is low contrast.

Those three place you in one of the twelve seasons. Depth and contrast are the easy ones to judge. Undertone is where almost everyone goes wrong, which is why most of the setup below exists to protect it.

Step 1: the setup

Find a window and face it. Daytime, indirect light. North-facing if you are in the northern hemisphere, south-facing in the southern. Stand facing the window so the light falls on your face, never with your back to it.

Turn off every artificial light in the room. This is the step people skip and it does more damage than all the others combined. A lamp on while you are near a window means two different light sources at two different color temperatures, and there is no single correction that fixes both.

Avoid direct midday sun and golden hour. Direct sun blows out the highlights and destroys the subtle information in your skin. The hour after sunrise and before sunset is around 3,000 K, which is very orange, and makes literally everybody read warm. Mid-morning or mid-afternoon, indirect.

Take everything off your face. All makeup. Foundation is engineered to change your apparent undertone; bronzer and blush do the same thing more obviously. Push your hair back off your cheeks so it is not bouncing its own color onto your skin.

Wear white or light grey, and clear the background. A red door or a bright green plant behind you will skew your phone's white balance and push your skin the opposite way.

Step 2: take the photo

Front camera, no filters, no portrait mode, no beauty settings. Check your camera app's settings specifically, several phones apply skin smoothing by default and do not advertise it.

Hold the phone at eye level, roughly arm's length, face filling a good part of the frame. Tap on your cheek to set the exposure there rather than on the background.

Take three or four. Do not pick the most flattering one. Pick the one where the light is most even across your face.

Step 3: read the three properties

Depth. Look at the photo squinting, or convert it to black and white on your phone. Does your overall coloring read as light, medium or deep? Black and white is genuinely useful here because it strips out hue and leaves only value, which is exactly what depth is.

Contrast. Still in black and white. How big is the jump between your hair and your skin? If they nearly merge into one tone, you are low contrast. If it looks like a graphic novel panel, you are high contrast. Most people are somewhere in between and that is fine, medium contrast is the most common result.

Undertone. The hardest one. Three checks, in order of reliability:

1. The white test. Hold a piece of pure white paper next to your face in the photo. Against true white, warm skin looks slightly yellow or golden, cool skin looks slightly pink or grey. This is more reliable than the vein test because you have a genuine neutral reference in frame.
2. Gold versus silver. Hold each next to your face in the same light. One will make your skin look brighter and more even; the other will make you look tired. Gold suggests warm, silver suggests cool. If both look fine, that is a real answer and it means neutral.
3. The vein test. Look at the inside of your wrist in daylight. Greenish reads warm, bluish or purple reads cool. It is the most quoted test and the least reliable, because vein color is affected by how deep the vessels sit under your skin, not only by undertone. Treat it as a tiebreaker, never as the answer.

Step 4: check your own answer

The one thing amateurs skip and professionals do not.

Repeat it on a different day. Same setup, new photo. If you land on the same season twice, you have a reading you can trust. If it flips between warm and cool, your lighting was not controlled and you should redo the setup rather than pick your favorite.

Sanity check it against your wardrobe. Which colors do you already own that people compliment you in? Which things hang unworn because you look washed out? Your own history is real evidence and it is free.

Be suspicious of any result you desperately wanted. People talk themselves into being a Bright Winter with impressive regularity.

Where the manual method runs out

Doing this by eye works reasonably well for depth and contrast. It is genuinely difficult for undertone, because you are trying to detect a shift of a few percent in the red-yellow balance of your skin, by eye, from memory, against a shifting reference.

That is a measurement problem, and it is the part worth handing to something that measures.

Our free color analysis tool does the same three readings from the photo you just took. It corrects for the light source first, then measures your undertone, depth and contrast from your actual skin pixels and returns your season with five colors to start from. It runs the same AI engine as the app: your photo is resized in your browser and sent over an encrypted connection to Google's Gemini API for the analysis. Neither we nor our Worker write it to storage. If the service is unreachable, the analysis runs entirely in your browser instead. If the lighting was not good enough for a clean read, it says so instead of guessing.

Do it both ways. If the manual read and the measured read agree, you are done.

Common questions