There's no formula to hack
A live face-duel score is a black box from the outside — the weighting between measurements, and whatever else the underlying model considers, isn't disclosed. That means "improving your score" can't mean reverse-engineering an exact formula. It means treating the reading like any camera-sensitive measurement: identify what you can actually change, change one thing at a time, and see if the number moves in a direction that holds up across more than one attempt.
Fast lever: camera and presentation
This is the lever that changes fastest, because it changes what the camera sees rather than your face itself. Distance, angle, lighting, and expression can each swing a reading meaningfully between two photos of the same person. Run the four-point practice checklist to work through each variable in order, or read the full practice guide for the reasoning behind each one.
Slower lever: know which measurement is actually moving
Once camera variables are controlled, the six geometry ratios behind a composite score respond differently to different things — some are lighting-sensitive, others angle-sensitive, others closer to a fixed structural trait. Reading the measurement-specific guides tells you which lever you're actually pulling before you spend time on it: jaw ratio, cheekbone ratio, canthal tilt, facial symmetry, midface ratio, and nose bridge ratio.
Confirm it worked instead of assuming it did
A single higher reading after a change could be the change working, or it could be the same noise that makes scores swing between scans in the first place. The only way to tell the difference is more than one data point: save a baseline, change one variable, save again, and read the pattern across saved sessions rather than trusting any single scan. Once a reading is stable, see what tier the composite actually lands in.
How Mogalytics helps with this
Mogalytics reads the same six camera-visible geometry ratios live from your own camera, entirely on-device, settled over a 30-frame trimmed-mean window so a blink or turn doesn't define a saved reading. It's built for exactly this kind of before/after test — nothing about it — camera frames, landmarks, or scores — ever leaves your browser.
Related reads
For the pre-duel camera checklist, see the Omoggle practice test; for the full variable-by-variable reasoning, see the practice guide; for why readings swing between attempts, see the accuracy breakdown.