Free · runs on your device

On-device skin analysis from one photo

This reads four things a selfie can actually show: how red the cheeks are, how even the tone is, how much shine is in the highlights, and how much local texture the light picks up.

It runs in the browser after a face model file loads. Your photo stays on the device. For a history of scans, use the AISkincare app.

Cost
Free
Time
Under a minute
Signup
None
Data
Stays on device

Your photo

Add a photo to measure

Face the camera straight on, hair off the face, even light, no filter.

or drag a file here
  1. 1Add your photoCamera or a file. Either stays on this device.
  2. 2Mapped in your browser478 landmarks find the zones. The picture is not uploaded.
  3. 3Your measured resultNumbers from this photo, drawn on the picture.

Add a photo. It stays in this browser.

  • On-device, never uploaded
  • 478 landmarks
  • Free, no signup

In the app

Want a record of ai skin analysis, not a one-off?

AI skin analysis on this page is a single reading. The app saves a scan and the routine that followed it, so the next photo has a baseline.

  • Six saved scores: hydration, texture, tone, acne, pores, and redness
  • A morning and evening routine built from the scan
  • The same face, week after week, so a change shows up as a trend
QR code that opens the AISkincare App Store listing on iPhone and Google Play on Android

Scan with your phone. iPhone opens the App Store. Android opens Google Play.

The science

What ai skin analysis is, and how this page works

  1. 01

    A measurement of the surface, taken at one moment

    Skin analysis turns what a face looks like into numbers you can compare later. A dermatologist does a version of that by eye and by touch. A clinic camera does it with fixed light. This page does it in software: find the face, sample skin pixels, and describe their color and texture. The output belongs to the moment the shutter closed. That is the useful part, and it is the whole part.

  2. 02

    The entire read stays in the browser

    A MediaPipe face landmark model places 478 points so the forehead, cheeks, nose, and under-eye skin can be found no matter where you sit in the frame. Those patches are drawn to a canvas on your device and converted from screen RGB into Lab, where lightness (L*), the red-green axis (a*), and the yellow-blue axis (b*) can be read apart. Local variation in lightness adds the texture cue. The photo is not uploaded. When you leave, the canvas is discarded.

  3. 03

    Four signals, and a fuller record in the app

    Four surface signals from the cheeks and forehead: redness, tone evenness, shine, and texture contrast. Each one is a description of this photo, scored so you can compare a later photo taken the same way.

    After 478 landmarks fix the zones, small skin patches are converted from screen RGB into Lab. Redness follows the a* axis (green to red). Evenness is how much L* (lightness) disagrees between cheek patches. Shine is the share of pixels that are both light and low in color. Texture contrast is local lightness variation inside the patch. A warm bulb, a side window, or a beauty filter moves those numbers without your skin having changed.

    The AISkincare app keeps a scan: hydration, texture, tone, acne, pores, and redness, then builds a morning and evening routine from those scores. This page answers once and does not store the photo.

  4. 04

    Light is part of every number

    This is a cosmetic reading of one photo on your device. Lighting, lens, and angle change the numbers. It is not a diagnosis or a clinical measurement.

    Warm bulbs push redness. Side light invents unevenness. A beauty filter deletes texture before the sample exists. The score describes your face relative to itself in this room, not against a population, and not against the face you will have in a different window tomorrow.

What ai skin analysis reports, and what the app adds

Redness rises when the cheek patches carry more a* than a neutral reading. A flush, lingering post-blemish color, and a warm lamp all do that, so the number is how red the surface looks now. Tone evenness falls when one cheek's lightness disagrees with the other, which happens with patchy pigment and also with a window on only one side. Shine is highlight on the cheek, the quick stand-in for a slick surface. Texture contrast rises with roughness, dryness, and a harsh sensor.

The app's advantage is the second scan. Hydration, texture, tone, acne, pores, and redness are stored with the routine you were following, so a change has a before. This page cannot show that line. It can give you today's point.

A phone read and a clinic camera are different instruments

Clinic systems hold the face still, standardize the lamps, and often shoot more than one pass: ordinary light, a cross-polarized pass that cuts shine so redness and pigment separate, and sometimes ultraviolet that shows what has not surfaced yet. They may also place you next to other people of a similar age. A phone in a bedroom has none of those controls, so it sees less and repeats less exactly.

What the phone has is repetition. It costs nothing and you can run it every week in the same window. The clinic answers a deeper question a few times a year. The app answers whether the routine is moving the six scores between those visits. Neither one diagnoses. A new, growing, bleeding, or non-healing mark needs a dermatologist regardless of either report.

How to take a photo this page can trust

Face a window in the day, with the sun not on your skin. Overhead bathroom lights dig shadows that score as unevenness. Evening bulbs push red and yellow. Flash paints highlights that score as shine. Hold the phone at eye level, about an arm's length away, and look straight into the lens with a relaxed face.

Clear hair off the forehead and cheeks. Skip makeup, skip a just-applied moisturizer, and turn off beauty mode and any smoothing filter. Then repeat those choices next time. A single score mixes your skin with your lighting. Only a repeated setup lets a change in the number mean a change in the skin.

Read the slope, not one afternoon

Sleep, a cycle, a cold walk, exercise, alcohol, and a new detergent all show on the surface within a day. Two readings a week apart mostly show that noise. Four to six weeks of photos in the same window, once or twice a week, is where a direction shows up: redness easing after you stopped a harsh acid, texture settling on a retinoid, shine unchanged despite a new cleanser.

This page is the free snapshot. Guided framing and a history live in the app, which is why the QR code under the tool opens the App Store on an iPhone and Google Play on an Android phone.

What ai skin analysis cannot tell you

It cannot see oil under the surface, water in the barrier, or a count of individual blemishes. It cannot separate a flush from inflammation. Those are the jobs of a saved app scan over weeks, and of a clinician when something hurts or is changing.

Rosacea, eczema, a plain flush, and a fungal-pattern breakout can all look red in a photo, and they are not the same decision. A dark spot can be a freckle, a mark left after a blemish, or something that needs a dermatologist, and no color average separates those. If a spot is new, changing, bleeding, or not healing, book the visit and do not wait for a better score.

How to use the result

  1. Use a photo taken by the rear camera, or by someone else, at arm's length.
  2. Face the light. Do not use a portrait filter.
  3. Keep hair off the forehead if you want the outline and thirds to settle.

Where to go from here

If one part of the result stood out, open the tool that asks only that question. Each one states its limit in the same place, and each one links back to a guide when a longer explanation helps.

This is a cosmetic reading of one photo on your device. Lighting, lens, and angle change the numbers. It is not a diagnosis or a clinical measurement.

FAQ

Questions about ai skin analysis

How should I use the ai skin analysis?

Treat a high redness or texture number as a reason to simplify the routine, not as an acne grade.

Does this page upload my photo?

No. The picture is measured in your browser. A face-model file is downloaded once so the measurement can run on your device. The photo itself is not sent anywhere.

What if the number looks wrong?

Retake at eye level, in window light, with a neutral expression. A low camera and a beauty filter will move every result.

AISkincare for iOS and Android

AI skin analysis reads one photo. The app keeps the scores.

This page runs ai skin analysis in your browser. The picture stays on your device. Open the app when you want that concern saved as hydration, texture, tone, acne, pores, and redness, then turned into a morning and evening routine.