Light: the biggest variable by far
Lighting changes scan results more than any other factor, so it earns the most care. Face a window with soft daylight, or a diffuse lamp placed straight ahead of you — the goal is even, frontal light that reaches both sides of the face equally. Overhead bathroom LEDs are the classic mistake: they throw shadows down the face, carving fake hollows under the eyes and jaw that a model can read as texture and softness. Strong side light is the other trap, splitting the face into a bright half and a shadowed half, which distorts how symmetry and contour read. Avoid backlighting entirely — a window behind you silhouettes the face and starves the scan of detail. Time of day matters for consistency too: morning window light and evening lamp light are different instruments, so pick one condition you can reproduce and scan in it every time.
Angle and distance
Hold the camera at eye level, face it straight on, and keep the distance consistent — roughly arm’s length, or a fixed spot if you prop the phone. The geometry here is unforgiving: a camera below eye level shortens and softens the jawline, one above stretches it and slims the face, and a few degrees of left-right rotation makes a symmetric face read as unbalanced. Close range adds lens distortion that enlarges whatever is nearest, usually the nose. None of these effects is subtle at scoring resolution, and all of them are invisible to you in the moment — which is why the fix is procedural rather than judgemental: same height, same distance, same head-on framing, every single scan. If your scores swing wildly between scans taken days apart, the angle is the first suspect, well before your face.
Face presentation
The scan needs to see the landmarks it measures, so clear the obstructions: hair off the forehead and away from the jawline, glasses off if you can, no hat. Expression should be neutral and repeatable — a slight smile in one scan and a tired scowl in the next will move symmetry and contour readings for reasons that have nothing to do with progress. No beauty filters, ever; a smoothed selfie produces a skin score for the software, and the routine generated from it aims at problems you do not have while missing ones you do. Makeup is a choice with a consequence: bare-face scans are the honest track for Skin Quality trends, so if you wear makeup in scans, keep it consistent, and know that you are tracking your presented face rather than your skin itself.
Make it a protocol, not a decision
The difference between people whose scan history means something and people whose history is noise is not diligence in the moment — it is that the first group decided the conditions once and stopped deciding. Write your protocol down mentally as one sentence: “weekend morning, facing the bedroom window, phone propped at eye level on the shelf, bare face, neutral expression.” Then never renegotiate it, especially not on days you feel good — the temptation to take a flattering off-protocol scan is exactly how histories go bad. The rescan cooldown helps by spacing scans out, and your ordinary cadence — weekly for skin experiments, monthly for body-composition work — does the rest. A protocol scan takes ninety seconds. Months later, when your before/after compare shows a climb you can trust completely, those ninety seconds are why.
