About Clad
Clad is a web-based AI personal color analysis. From a face photo, it estimates your season — Warm Spring, Cool Summer, Warm Autumn or Cool Winter — and any finer tendencies, then suggests makeup and fashion colors that tend to suit you. There is no app to install; you can start right away online.
What we help with
- • Estimating your main personal color type and warm/cool tendency
- • A palette of colors that suit you, plus colors to go easy on
- • Color guidance for lips, eyeshadow, blush and clothing
- • A light read on your tone direction before an in-person session
The principles we hold to
Personal color is not a scientific measurement but a style framework about which colors suit you. Clad tries not to overstate its results, is upfront about the limits of online analysis (lighting, editing, displays), and offers the result as a reference guide to help with style choices.
Who operates Clad?
Clad is developed and maintained by LumX Product Studio in Korea. The Clad Editorial Team reviews the sources, dates, and claims in product guides, FAQ, and comparisons. Send product, photo deletion, privacy, or correction questions to admin@lumx.im.
How is a result produced?
The service checks photo format and whether the input is usable, then structures skin-color tendency, brightness, and contrast observations from the needed face region. A versioned AI response is validated against a server schema to estimate season and subtype, and the validated result is connected to palettes and makeup and fashion guidance. Questionnaire answers add context to the explanation but do not arbitrarily override the photo-based type.
How do we handle accuracy and validation?
Model, prompt, preprocessing, and response-contract changes are compared as versions on evaluation data separated from production uploads. Checks cover seasonal results, retake decisions, consistency, failures, and latency. Result text is separately checked for fixed facts, prohibited claims, locale correctness, and safe fallbacks. We do not promote an unsupported single accuracy percentage before representative validation evidence can be published with it. Read the full analysis methodology.