The background choice, decided by consistency

People ask which background colour scores better, expecting the answer to be black, or white, or some specific shade in between. The answer is that colour is not the variable that matters. Reproducibility is, and a background you can set up identically every time beats a "better" one you can only approximate.

Why the colour question is the wrong question

A plain, matte background exists to keep clutter out of the frame and to give the rubric a clean field to read the subject against, not to add its own visual quality to the score. Once a background is plain and consistent, its specific hue does very little on its own - the model is reading the subject, not grading the backdrop. That framing follows directly from standardising a submission rather than flattering it: the background is one of the variables that is not the subject, and the job is to fix it, not optimise it.

Where colour does matter: exposure

The one place background colour genuinely interferes is auto-exposure. Cameras that meter automatically off the whole frame will read a very dark background and brighten the exposure to compensate, or read a very bright background and darken it, shifting how the actual subject is exposed even though nothing about the subject changed. A pure black backdrop and a pure white backdrop are the two ends most likely to trigger this, which means either extreme needs a manually locked exposure setting to stay neutral - the same discipline covered for exposure generally, where the rule is fix it, do not let the camera chase it.

A mid-grey or muted backdrop sidesteps the problem by not pulling the meter toward either extreme, which is one reason it shows up often in setups built for consistency. It is not a requirement. Black and white both work, provided exposure is locked rather than left on auto.

The practical case for each

Black has one advantage: it hides seams, folds and minor imperfections in the backdrop material itself, so a cheap sheet of fabric or paper reads as clean even when it is not perfectly smooth. It also tends to make the subject stand out with more contrast, which some people find easier to line up consistently in a viewfinder.

White has the opposite advantage: it is easier to source consistently, since printer paper, a wall, or a bedsheet are all close enough to white to start with, while a true, deep black backdrop that does not pick up reflections takes more deliberate setup. White also shows dust, shadows and creases more readily, which is a downside for looks but an upside for consistency, because a flaw that is visible is a flaw you notice and fix before it drifts unnoticed between sessions.

Neither advantage is about the score. Both are about how easy each colour is to keep identical across many sessions, which is the only property that matters here.

There is also a practical middle path worth naming: a backdrop that is easy to buy a second identical piece of, in case the first one wears out, tears or gets thrown away. A specific fabric or paper roll that can be repurchased by name is more durable as a long-term standard than a wall or a bedsheet that cannot be replaced exactly, and durability of the standard itself is worth weighing alongside the colour question.

What actually breaks comparability

Clutter, changing furniture behind the subject, a background that shifts between beige one week and patterned the next - not because any of those are inherently worse to a rubric, but because none of them repeat. How a model processes the full frame does not selectively ignore the background; a busy one is read alongside the subject rather than filtered out, which is exactly why a plain and constant one removes a source of variance rather than a source of appeal. The effect is large in published tests: Xiao, Engstrom, Ilyas and Madry (2020) found image-recognition models can reach non-trivial accuracy from the background alone, and misclassify images with a correctly classified foreground "up to 87.5% of the time" when the background is adversarially chosen.

This has nothing to do with what makes a backdrop suit a physical reading, where a measurement setup cares about a fixed reference object being visible in frame rather than about colour or clutter at all, and nothing to do with what a human viewer might find visually appealing in a backdrop, which is a taste question this site does not have a stake in.

Pick a colour, matte finish, lock the exposure, and use it every time. That is the entire decision, and it is the same decision whichever result you send to Rate Cock or anywhere else.