AI face analysis
One selfie, four honest reads from a free AI face analysis, and the next step for each.
Nothing is uploaded; the face analyzer runs inside this tab.
Cross-reading
How the four face analysis reads explain each other
An AI face analysis is a single-photo read of four things — face shape, facial balance, apparent age and skin tone — produced by a landmark model that runs inside your browser, so the selfie is never uploaded. Taken alone, each of those four numbers is a bit thin. Taken together they check one another, which is the one thing an AI face analysis hub can do that four separate tools cannot, and it is where the result stops being trivia and starts being useful.
Age and skin tone read the same pixels
In an AI face analysis, age and skin tone share the same evidence: contrast between the eyes, brows and lips and the skin sampled four times. Those reads share the same pixels, so if age lands higher than expected while the colour panel flags a shadowed sample, both are describing the light in the room, not you. Move to a window under even daylight; if both shift together, you have diagnosed the light rather than learned something new about yourself.
Shape and balance trace different points
A face analysis uses different points for shape and balance. One traces the jaw, cheeks and temples; the other checks the hairline, brow, nose and chin. Hair or a hood can hide the first set without changing the second, so an odd outline beside a confident balance read usually means an obstruction. Check it with a full shape reading before committing to a new cut. Symmetry depends on all of them, which is why the page refuses that verdict when your head is turned past about fifteen degrees.
Evidence
One photo, four kinds of evidence a face analyzer weighs
A face analysis uses one photo, but each read trusts a different part of it. A picture does not fail uniformly; angle, light, framing and obstruction distort specific evidence in predictable ways. That is why a face analysis result can look reliable in one panel and shaky in the next. The table separates those failure modes instead of collapsing them into a single photo-quality warning. Check it before deciding the result is wrong.
| What is off in the photo | Shape | Balance | Age | Skin tone |
|---|---|---|---|---|
| Your head is turned a few degrees | holds | distorted | holds | holds |
| Warm lamplight instead of daylight | holds | holds | shifts | shifts |
| Hair falling across the jaw | hidden | partial | holds | holds |
| Phone held low, close to your face | distorted | distorted | shifts | holds |
Read down a column for each tool’s blind spot, or across a row for the fix. One retake by a window, with the phone at eye level, repairs three rows at once. It is the cheapest improvement to any face analysis: thirty seconds beats re-running the same photo. A changed setup is more informative than another identical attempt.
Over time
The face analysis reads that change, and the ones that do not
A face analysis is a snapshot, and these four reads sit on very different timescales. Knowing which is which tells you where effort is worth spending. A useful comparison respects those timescales and asks whether the photo changed or the underlying signal did under matching conditions.
Your outline shape is effectively fixed
A face analysis only needs to name that underlying outline once. Bone structure settles in your twenties and barely moves, so spend attention on what the result implies rather than measuring it again.
Balance is fixed, but its reading is not
Two face analysis reads a week apart that differ noticeably are two different photos, not two different faces. A level, front-on retake is the best check.
Apparent age moves the most, and moves back
This face analysis read responds to hair, collar colour, sleep and light, so it is the one worth repeating. That makes it an experiment, not an identity label.
Skin tone drifts seasonally
This face analysis read shifts in depth with sun and season while its underlying character holds. Compare it under the same daylight before treating different swatches as meaningful.
That split is the reason for keeping a face analysis. Save this face analysis read, change one variable — a neckline colour, hair shade or a draped swatch under the chin — then run it again and compare only what moved. Holding every other condition steady turns the second photo into evidence instead of another guess. The step-by-step path orders those experiments so each informs the next. If colour is the panel you want to pursue, the full colour analysis turns three swatches into a season and a palette.
Questions
Face analysis, honestly answered
Does this face analysis give a beauty score?
No. This page reports shape, balance, apparent age and three skin tones, not a beauty score. A single rating belongs somewhere built to handle it kindly: our attractiveness test leads with strengths and keeps the number optional. If a score is what you came for, open that tool; this face analysis will not give you one.
How is this face analyzer different from the face shape, age guesser and attractiveness tools?
The specialist tools answer one question each in depth. A face analysis answers all four at once, lightly, and shows where every answer came from on your photo. Think of it as the contact sheet: it flags which read is worth pursuing, then hands you to the page with the deeper styling advice, age levers or proportion detail.
What if two of the reads seem to disagree?
Disagreement is usually information. An older age read beside a shadow-heavy skin sample points to the light in the room. A strange shape result beside weak colour confidence usually means hair or a hood covers the jaw. Trust the read whose evidence the photo supports; the table above shows which. A face analysis that disagrees with itself is usually describing your photo, not your face.
Do you measure canthal tilt, FWHR or the other looksmaxxing numbers?
No. Those measurements are real geometry, but ranking people against a supposed ideal is the opposite of this site's purpose. A front-facing phone photo also cannot support the confidence attached to those numbers. This face analysis reports proportions as proportions, without a leaderboard.
What actually leaves your browser during a face analysis?
Your photo does not. MediaPipe and the age model process it inside this tab. Only the model files download once; after that, a face analysis needs no network at all. The tab temporarily holds the read and a smaller copy so links do not erase your result, and closing it clears both. Saving to your account stores only the four conclusions and skin hex codes; the image is never saved. The one opt-in exception is the haircut render under the read: it sends a small square of the photo to our image model once, only when you tap it, and does not keep it.
Next
Where your face analysis goes next
A face analysis is a starting point, not a verdict; the face analyzer above only tells you which face analysis read is worth pursuing, and each panel opens onto the page that takes that question seriously.
Face shape detector
Seven shapes, their look-alikes, and the lines each wears well.
Face attractiveness test
Your strengths, optional score, and the levers that move a look.
Age guesser
Five photo cues explain the read; three can change.
AI colour analysis
Turn three skin swatches into your season and wearable palette.
AI hairstyle changer
Try shape-aware cuts on your photo before visiting a stylist.
Hair type scanner for men
Read curl pattern, density, and what your barber needs to hear.
Beauty Path · 5 steps
Had fun? There’s a route through all of this.
Five steps, in the order that actually matters: your colors first, then makeup, hair, and shape. Each one takes a few minutes and leaves you with something you can use the next time you get dressed.