One portrait,
nine observations.
Refine is a beauty advisor, not a scanner. It looks at a single photo the way a good makeup artist would across a table, notes what a product could visibly help with, and stops there.
1. You upload a portrait
Any clear, front-facing photo works. Natural light and no heavy filters give the most useful read. The image is resized in your browser before anything leaves your device, so a phone photo of several megabytes becomes a few hundred kilobytes.
You can drag a file in, choose one, or paste an image straight from the clipboard.
2. Your device finds the face
A small face-landmark model runs entirely in your browser and locates 478 points on the face: the corners of the eyes, the edge of the lips, the line of the jaw and so on. These are what let the result screen draw a line from each recommendation to the exact spot it refers to.
If no face is found, nothing is sent anywhere and you are asked for a clearer photo. Group photos are read on the first face detected.
3. An AI beauty advisor looks at nine zones
The reduced image is sent to a vision model with a fixed brief: describe only cosmetic surface features, in nine zones, kindly and specifically. The zones are under-eyes, brows, lashes, lips, skin, cheeks, forehead, nose and jawline.
The brief explicitly forbids comment on age, weight, ethnicity, attractiveness or anything that looks medical. If the model is unsure about a zone, it says so or leaves it out rather than guessing.
4. Each observation is paired with one product
The advisor chooses from a short curated list of widely available skincare and makeup items rather than the whole of Amazon. Every product on the list was picked because it is well reviewed, easy to find, and does one clear thing. Stronger observations lean toward treatments; lighter ones toward everyday products.
Links are Amazon affiliate links. As an Amazon Associate we earn from qualifying purchases. The price you pay is unchanged, and the advisor never sees commission data, so it has no reason to prefer one product over another.
5. Nothing is kept
The photo is held in memory for the length of one request and then discarded. It is not written to disk, logged, or used to train anything. If you choose to email yourself the list, only the text of the recommendations is sent. Read the privacy page for the full detail.
Refine reads your features from one portrait and suggests a product for each. Nothing is stored.
Upload a portrait