Why the capture path changes the image

A mirror shot and a front-camera shot are different submissions from a direct rear-camera photo, even thirty seconds apart: the mirror roughly doubles the effective distance and flips the image, and the front camera brings its own lens and processing. A rating tool cannot tell, because both files look like ordinary photographs.

What a mirror actually adds

A mirror shot is not a direct photograph of the subject. It is a photograph of a reflection, taken from whatever distance you are standing from the glass, of a reflection that is itself the same distance again behind the mirror surface. That doubles the effective distance between lens and subject compared with a direct shot taken from where you are standing, which changes the framing and the degree of wide-angle distortion at the edges of the frame. Camera distance is already the single largest source of score movement that has nothing to do with the subject, and a mirror shot moves that variable in a way that is easy to miss because you never measured "distance to the glass plus distance behind it" - you just stood where you always stand.

Mirror glass is also rarely perfectly neutral. Cheap mirrors have a faint cast, and even good ones lose a fraction of a stop of light to the reflection, which the camera's exposure and colour systems then compensate for automatically. And a mirror photograph is flipped left to right relative to how a direct camera would have captured the same scene - not something that changes a numeric score, but a reason a mirror series and a direct series should never be treated as continuous with each other in the first place.

What a front camera actually is

Switching to the front camera is not just moving the phone. On almost every current phone, the front camera is a physically different sensor and lens from the rear ones, usually a shorter focal length and a wider field of view than the rear main lens, which changes how proportion renders at a given distance in the same way focal length does generally - the front camera is simply one fixed instance of that variable rather than a choice among several.

Front cameras are also usually used at arm's length, which is close enough for perspective to matter. Ward and colleagues (2018) calculated that a face photographed at 12 inches has its nose enlarged by 30% in men and 29% in women compared with an undistorted projection, and that at 5 feet there is essentially no difference.

Front cameras also carry their own image processing pipeline, separate from the rear one, and on many phones that pipeline applies some degree of smoothing or "beauty" correction by default, sometimes without an obvious toggle. That processing is not something you dialled in. It is a manufacturer default running on every front-camera shot whether you want it or not, and it changes fine detail before the file ever reaches a rating tool.

Why mixing them breaks a series

A submission series is comparable only when the capture path stays fixed, and both of these count as capture path, not as incidental variation. A baseline taken with the rear camera, direct, and a follow-up taken in a mirror with the front camera, differ on at least three axes at once: effective distance, lens and field of view, and processing pipeline. If the score moves, there is no way to say which of the three did it, or whether the subject changed at all. That is the same failure mode as changing two things at once between any pair of submissions - except here the two changes are hiding inside what looks like a single decision, "which camera to use."

A tool that shows the breakdown behind a result, the way Rate Cock exposes per-axis scores on public entries, at least gives you a chance of noticing that one axis moved oddly after a camera switch - a total-only tool gives you nothing to check the swap against.

Neither path is the wrong one to standardise on. A rear-camera direct shot is generally the more controllable choice, because it avoids the mirror's added distance and the front camera's default processing, but a front-camera or mirror series can be just as usable as long as it stays a series - the same setup, repeated, rather than swapped in and out session to session. The failure is not choosing one method. It is treating the two as interchangeable because they both technically produced a photo of the subject, which is the same mistake as assuming a professional setup and a phone camera produce comparable results just because both are photographs - the equipment matters less than whether it stayed the same. None of this applies to a measurement taken with a tape, which has no lens, no mirror and no processing pipeline to vary in the first place - a reminder that the capture-path problem is specific to photographs, not to the underlying property they are trying to show.

What to record

Note which camera and which path, alongside the rest of the log. "Front camera, mirror, bathroom light" is a condition exactly like distance or angle, and it should sit on the same line in a record of the conditions as everything else that has to hold still for a result to mean anything on a second attempt. The rendering differences between a rear lens, a front lens and a lens seen through glass are a subset of a wider question - what a camera's optics do to proportion at a given setup - and the honest fix is the same one that applies everywhere else in a protocol: pick one path, use it every time, and treat a switch between them as starting a new series rather than continuing the old one.