About personalcolor.co

A few years ago, a friend came back from a 퍼스널 컬러 (personal color) consultation in Seoul. She'd paid ₩99,000 for a 90-minute session, and she was glowing — not from the analysis itself, but from the confidence that came with finally knowing which colors were hers.

I was curious. I started reading about the 4-season color system, the difference between warm and cool undertones, why certain colors make some people look tired and others look alive. The more I read, the more I realized this wasn't magic — it was color science. LAB color space, chromatic temperature, reflectance values. Things a developer could work with.

So I asked the obvious question: why does this cost ₩99,000 and require an appointment? The answer was mostly "because that's how it's always been done." The analysis itself — sampling skin tone, measuring undertone warmth and brightness, mapping to a seasonal type — is a deterministic process. It can be automated.

I spent several weekends building the first version. I learned face-api.js, read papers on skin tone classification, and ran the tool on every photo of friends and family I could find. Most results matched what professional consultants had told them. Some didn't — and that taught me about the limitations: lighting conditions, camera white balance, heavy makeup. I documented all of it.

The site launched quietly. Then K-Beauty content creators started sharing it. Then international users found the English version. Then people started asking for more — makeup guides, color palettes, and eventually the 관상 (face reading) feature that pairs traditional Korean physiognomy with modern first-impression psychology.

Today personalcolor.co is still a one-person project. No VC funding, no team, no office. Just a developer who got curious about color theory and decided to make something useful out of it.

Personal color analysis is rooted in color theory developed in the 1980s by Carole Jackson (Color Me Beautiful) and later refined by Korean beauty industry practitioners into the 8-subtype system widely used today. The core principle: everyone has an underlying skin undertone (warm/cool) and a brightness level (light/deep), and certain colors harmonize with those properties while others clash.

Our algorithm works in LAB color space rather than RGB because LAB separates luminance (L) from color information (A = green-red axis, B = blue-yellow axis), which more closely matches human color perception. Skin tone warmth maps to the B axis; brightness maps to L. This makes classification more stable across different camera sensors and lighting conditions than a simple RGB comparison.

The 관상 (face reading) feature draws on classical Korean physiognomy — a system of reading personality and disposition from facial features that has been refined over centuries — and cross-references it with modern social psychology research on first impressions and facial feature perception. It is designed as an engaging, reflective experience rather than a predictive tool.