Drop one photo
Your free upload. Front-facing, even light, no filters — that is all the algorithm needs. Everything is read in the browser.
One photo in, one full facial breakdown out — cheekbones, jawline, eyes, symmetry, proportions, lips and skin. The landmark model runs in your browser, so your face never leaves this device.
Drop your face here
or click to browse · drag & drop · paste with ⌘V
JPG / PNG / WEBP · up to 10 MB
No photo handy?
Unlimited scans · analysis happens locally, nothing is uploaded.
Scanningface
—
—
—
—
—
No face found
Try a front-facing photo in decent light — one face, no sunglasses, no heavy crop — then try again.
Runs on a 478-point face-landmark model in WebAssembly · no server, no account, no storage.
Your free upload. Front-facing, even light, no filters — that is all the algorithm needs. Everything is read in the browser.
MediaPipe pins 478 points across your face while face-api independently places 68 landmarks, reads expression and estimates age. Both analyses are averaged into one score.
Seven weighted traits roll into one PSL number with a grade, a verdict and a downloadable scorecard. Straight from your pixels.
Each trait is a real geometric measurement, mapped onto a 0–100 curve with its own ideal range, then weighted into your PSL score.
No. Both models, their runtimes and your pixels all stay inside this tab. There is no backend, no account and no storage — the site is a single folder you can host anywhere, or run offline.
MediaPipe's 478-point landmarker is the geometric backbone, but any single model can be wrong. face-api is a completely different stack (TFJS + TinyFaceDetector + 68 landmarks) that independently finds the face, so the two act as a cross-check — and its extra outputs (expression, age) show up in your readouts.
Personal Symmetry Level: seven weighted facial metrics (cheekbones 15, jawline 15, eyes 15, symmetry 15, proportions 15, skin 15, lips 10) combined into a single 0–100 number with a letter grade.
Landmark geometry only. Cheekbone prominence comes from the concavity under the widest row of the face outline; jawline from the gonial angle and jaw-to-chin taper; eyes from interocular spacing, openness and canthal tilt; symmetry from mean mirror deviation of all points; skin from texture statistics sampled inside the outline.
No. It is entertainment built on real computer-vision measurements. Attraction is not a formula, and a photo is not a face — use it for fun, not for self-worth.