HOW WE ACTUALLY FIND YOUR COLORS

Your skin has a specific combination of melanin, hemoglobin, and carotene that determines its undertone — the subtle warmth or coolness beneath the surface. When you wear a color that harmonizes with that undertone, something interesting happens optically: the color reflects light that complements your skin, making it appear smoother, more even-toned, and more radiant.

When you wear a color that clashes with your undertone, the opposite happens. The reflected light emphasizes unevenness, makes shadows look deeper, and can make you look tired or washed out. This isn't subjective — it's how light interacts with pigment.

The mechanism behind this is called simultaneous contrast: your visual system processes color through opponent channels (red vs. green, and yellow vs. blue). A color near your face doesn't just sit there — it actively shifts how your brain reads everything next to it. Wear a cool tone and your warm skin reads even warmer. Wear a muddy tone and clear skin looks duller by contrast.

Professional image consultants have understood this for decades. They hold colored fabrics against your face (a process called “draping”) and observe how your skin responds. The right drape makes you glow. The wrong one drains you. It's immediately visible.

What we've done is translate that same observation into a digital process — using the same color science, applied computationally.

Is color analysis actually real? Read the full breakdown →

Most online quizzes give you one of four seasons: Spring, Summer, Autumn, or Winter. That's like saying your shoe size is “big” or “small.” It's technically directional but practically useless.

The professional system uses 12 seasons — three subtypes within each of the four families. The distinction matters because each subtype needs genuinely different colors. A Soft Summer (muted and gentle) looks terrible in the icy brights that make a True Winter glow. Grouping them both under “cool tones” is the reason most quiz results feel wrong.

The 12 seasons are defined by three measurable dimensions of your natural coloring:

Every person lands somewhere in this three-dimensional space. The 12 seasons are regions within that space — clusters where specific color recommendations work consistently. When we say “you're a True Autumn,” we're saying your temperature, value, and chroma values place you in a specific region where warm, rich, earthy colors harmonize with your natural coloring.

When you upload a selfie, our system doesn't just eyeball it. It runs a multi-stage pipeline that progressively transforms your photo into reliable color evidence.

One important caveat before we start: pixel colors in a photo are not ground truth. They're filtered by the camera's white balance, the color temperature of the light, and the background behind you. A selfie taken under warm indoor lighting can shift your apparent skin tone by an entire season. This is called metamerism— the same surface looks different under different light sources. It's why we ask for a photo in natural daylight, and why the paid tier re-analyzes your image with these factors in mind.

CIELAB (also called Lab) is the backbone of our analysis. It was developed by the International Commission on Illumination (CIE) to model human color perception. Unlike RGB, which was designed for screens, CIELAB was designed for eyes.

It has three axes:

• L*Lightness — 0 is black, 100 is white. Your skin's L* tells us your depth.
• a*The red-green axis. Positive = more red/pink. Negative = more green. Skin a* values help distinguish warm (redder) from cool (less red) undertones.
• b*The yellow-blue axis. Positive = more yellow. Negative = more blue. The ratio of b* to a* is one of the strongest signals for warm vs cool classification.

We also use the cylindrical form, LCh, which repackages a* and b* into Chroma (how vivid) and Hue (the color angle). This makes it much easier to describe someone's coloring naturally: “moderately light, warm-leaning hue, medium chroma” maps directly to LCh values.

Why not just use RGB? Because RGB distances don't correspond to perceptual differences. A shift of 20 units in the red channel might be barely noticeable in one part of the spectrum and dramatically obvious in another. CIELAB was specifically designed so that equal distances = equal perceived differences. That's why it's the standard in color science, printing, and professional image analysis.

Most color analysis apps sample a single pixel or compute one average from a region. Both approaches are fragile.

A single pixel might land on a highlight, a shadow, a blemish, or a spot where the lighting shifted. One pixel is not evidence — it's a lottery ticket.

A simple average is better but still problematic. Averaging collapses structure. If your cheek has warm midtone skin (the signal) plus cool shadows (noise) plus bright highlights (also noise), the average lands somewhere between all three — a color that doesn't actually exist on your face.

Our approach uses clustering. We sample hundreds of pixels from each region and group them into clusters by color similarity. The dominant cluster that isn't a shadow or specular highlight is your true skin color. This is the same principle behind K-means clustering in machine learning — find the natural groups in the data and use the biggest, most representative one.

In the free tier, you guide this process by tapping multiple spots on your skin, hair, and eyes. Each tap samples a small grid of pixels, and we accumulate and average across all your taps. More taps = more stable result.

In the paid tier, our system automates this at scale — sampling thousands of pixels from properly detected face regions and running the full clustering pipeline. That's why the paid analysis is more accurate: more data, better algorithms, same color science.

Professional consultants use their trained eye under controlled lighting — and they're very good at it. What we offer is the same color science foundation, applied computationally, at a fraction of the cost. We don't claim to replace the experience of a skilled consultant. We do claim to be dramatically more accurate than a vein-color quiz.

Each of these articles covers one part of the color science in full detail — the mechanism, the evidence, and what it means for how you get dressed.

The color science behind our analysis draws from established literature and standards:

• • CIELAB Color Space — CIE 15:2004, Colorimetry (International Commission on Illumination)
• • Munsell Color System — the perceptual color framework underlying seasonal analysis
• • CSS Color Module Level 4 — W3C specification for lab(), lch(), and color space handling
• • Seasonal Color Analysis — originated by Johannes Itten, refined by Carole Jackson (“Color Me Beautiful”, 1980), expanded to 12 seasons by professional image consultants
• • K-means Clustering — Lloyd, S.P. (1982), “Least squares quantization in PCM”
• • WCAG Contrast Guidelines — W3C Web Content Accessibility Guidelines for color output accessibility