LEVEL / preparing
Loading LEVEL.
What actually happens in between, including the step that stops the process rather than continuing it.
Before a camera opens: a date of birth, and where you live. Both decide whether a photograph may be collected at all, and both come before anything is taken rather than after a questionnaire.
Front neutral, front smiling, both three-quarters, both profiles, and the hairline. A face model runs in your browser and checks brightness, sharpness, distance, centring, tilt and angle before each frame is allowed to upload — a photograph that fails those checks never leaves your machine.
Each analysis category runs tests that were reviewed and released, and every finding carries a confidence value describing how strongly your photographs support that specific observation.
A finding below the confidence threshold is not shown with a caveat attached. It is withheld, and the released result states what was withheld and why. Uncertain output is a worse product than less output.
Five reversible presentation directions are generated from your own photograph. Each is checked by independent validators for identity similarity, landmark drift, protected-region change, skin-tone consistency and edit containment. A preview that drifted is regenerated or withheld, never shown.
The released findings and approved recommendations become a schedule of reversible actions sized to the time, budget and routine you gave, with weekly check-ins that compare your own photographs rather than anybody else's.
They are stored in private object storage that is never given a publicly addressable URL, and destroyed automatically after 365 days without a capture or a completed mission. You can delete them, and everything else held about you, immediately from your account. They are not used to train any model. The processors that receive one are named in the privacy policy.
The reviewed tests behind each category, their evidence requirements and stated limitations are published in the methodology registry, with the sources in the research library. A complete worked result is at examples, and the analysis itself is explained field by field at inside the analysis.
A confidence-aware presentation profile, five reversible directions and a twelve-week plan. No rating, no ranking, nothing surgical. Photographs are destroyed after twelve months and never train a model.