About
We built the face analysis tool we wanted to use
LooksLab is a small product made by people who got tired of the two kinds of face apps that already exist. The first kind dresses you up with a cartoon filter and calls it a day. The second kind hands back a number with no explanation, a stock photo of a model, and a push toward an in-app purchase. Neither tells you anything real.
Why this exists
People are genuinely curious about facial geometry. The research on harmony, angularity, sexual dimorphism, and structural correlates of health is decades old at this point, and most of it lives in cosmetic surgery textbooks and psychology papers rather than in any tool a normal person can use. We wanted to put a well-calibrated version of that literature into a clean web app, and we wanted to show our work while we did it.
What it does
You upload a front photo and a side photo. LooksLab runs an in-browser landmark detector, then sends the resulting landmark coordinates (not the photo) to our scoring server, which runs each measurement through a calibrated scoring curve. The output is four pillar scores (Harmony, Angularity, Dimorphism, Health Indicators) plus the full list of underlying measurements if you want to see the receipts. The whole flow takes about a minute on a normal phone or laptop.
The full breakdown of how the pipeline works is on the methodology page, where we keep the explanation honest about what it can and cannot tell you.
Who built it
LooksLab is built by a small team of software engineers and designers who care about the intersection of geometry, computer vision, and product craft. We started the project in 2025 and shipped the first public version in 2026. We are not a venture funded startup with a marketing budget; we are a few people with a side project that grew up.
What we believe
Three principles shape the product. The first is privacy. The landmark detector runs in your browser, our scoring server only ever sees landmark coordinates (not your photo), saved photos live in a Supabase Storage bucket behind row-level security, and every scan can be deleted from the My Scans page.
The second is transparency. We publish the methodology, the count of curves, and an honest list of the limits. If you want to know how a particular measurement is computed, the answer is on the methodology page rather than buried in a marketing promise.
The third is restraint. A score from one photograph is a snapshot, not a verdict. We try to write copy that respects that, and we resist the temptation to package a number into a personality claim. The geometry of a face is interesting on its own; it does not need to be inflated into something it is not.
Get in touch
Press, partnerships, feedback, and bug reports all go to hello@lookslab.app. A human reads each one, and we try to respond within a few days.