AI facial aesthetic analysis · 100% in your browser

Understand your face.
Then enhance it — without surgery.

Upload one clear, front-facing photo. We measure 18+ facial proportions and skin qualities against peer-reviewed research, score each region, and build you a personalized, non-surgical glow-up protocol. Your photo never leaves your device.

  • 🔒 On-device — no upload
  • 🧪 Citation-backed metrics
  • 🧭 Diagnostic, never judgmental
Eyes88
Skin71
Jaw82
Lips90
Sample region scores

Three steps

1

Upload a photo

A clear, front-facing, neutral-expression photo in even light. Processed locally.

2

Get your report

Region-by-region scores with the measurement, the normative range, and the citation behind each.

3

Follow your protocol

A prioritized, non-surgical plan: skincare, grooming, makeup, and facial-exercise actions ranked by impact.

Analyze my photo

Scores are calibrated to your age × sex cohort.
Sets sex-conditional normative ranges.

The science (and its limits)

Metrics are grounded in peer-reviewed facial-aesthetics research: Little, Jones & DeBruine (2011) on symmetry and averageness; Farkas (1994) anthropometric norms; Driessen et al. (2011) on canthal tilt; Naini & Gill (2017) clinical facial metrics; Fink, Grammer & Matts (2006) on skin homogeneity.

Calibrated, not absolute

Every value is compared to an age × sex cohort range — what's typical and what literature considers ideal are shown separately, never conflated.

Explainable

Each score expands to show the measurement, the normative band, and the citation. No black box.

Honest about uncertainty

Photo quality, pose and lighting widen confidence intervals. Beauty is subjective and cultural; these are proportion measures, not a verdict on worth.

Privacy

  • Nothing is uploaded. The analysis runs entirely in your browser via on-device machine learning. Your photo and results never touch a server.
  • Nothing is stored. Close the tab and the analysis is gone. There are no accounts, cookies, or trackers in this build.
  • The model files (facial-landmark detector) are fetched once from a public CDN; your image is never sent with them.