What changing your distance to the lens does to a result

Move the camera closer or further from the subject and the score moves with it, before anything about the subject has changed. Of every variable a submission carries, distance is the one that moves a result the most for the least effort - a step forward or back, nothing else touched.

Why it moves anything at all

Phone lenses are wide-angle by default, and a wide lens enlarges whatever is nearest to it relative to whatever sits further back, inside the same frame. Step closer and that effect gets stronger; step back and it eases off. The size of the effect has been measured on faces: Ward and colleagues (2018) calculated that a photo from 12 inches enlarges the nose by 30% in men and 29% in women against a distortion-free projection, and that at 5 feet there is "essentially no difference." It also changes judgements, not just geometry: Bryan, Perona and Adolphs (2012) photographed the same faces at 45 cm and 135 cm, and viewers rated the close shots lower on trustworthiness and attractiveness. Nothing about the subject changed between the two shots - only the geometry the lens used to record it. Why a model reads that geometry the way it does is a longer subject than this post needs; the short version is that the tool is scoring the image, and the image is what the lens produced, not what a tape would have found.

The direction, not the size

It is worth being precise about what distance does and does not tell you. It does not push a score up or down in a fixed direction - closer is not "worse" any more than further is "better". What it does is change the proportions the tool sees, and a rubric built around proportion will register that change as a real one, because as far as the tool is concerned it is real. Two photos of the same subject taken at two distances are, to the tool, two different submissions. That is the entire finding: distance is a variable, not a dial you can turn toward a higher number.

Why fixing it is the first rule

Standardising a submission means holding constant everything that is not the subject, and distance is the largest single item on that list. It is also the cheapest to fix. Angle needs a level surface or a steady hand, light needs a consistent source, but distance just needs a mark on the floor or a note in a file: "same spot, every time." Skip that one step and every later comparison - this month against last month, one tool against another - is comparing two unknown setups rather than one held-constant one. The full ranking of which variables move a score the most puts distance near the top for exactly this reason: high leverage, low cost to control.

What this is not

It is tempting to read "distance changes proportion" as a size claim, and it is not one. This site does not deal in centimetres, and neither does a camera - an actual length figure comes from a tape and a documented method, not from how close someone stood to a lens. That is a different property with its own method, and conflating the two is the single most common way a photo-based result gets over-read as a measurement. A rating tool sees proportion in a frame; it does not, and cannot, recover a length from it.

If the photo is going to a person instead

The same discipline holds even when a model is not doing the reading. Anyone sending material to a human judge is still handing over a photograph taken from some distance, and a viewer's impression of proportion is shaped by the same wide-angle enlargement a model would pick up on - the audience changes, the geometry does not.

Why "closer" and "further" are not two symmetrical mistakes

Standing too close is the more common error, and it is also the more damaging one for a reason that has nothing to do with how close is "too close." Near-lens enlargement scales non-linearly with distance - the closer the subject gets to the lens, the faster the enlargement effect grows for a given step forward, so a half-step correction near the lens changes proportion far more than the same half-step would further back. That is why "further back with a crop afterward" is not just a preference; it is a way of moving the working distance into the flatter part of that curve, where small positioning errors between sessions cost less. Standing too far away has its own cost - detail thins out and framing gets harder to repeat precisely - but it does not carry the same non-linear penalty, which is why it is the safer direction to lean when in doubt.

A quick way to tell if distance moved between two results

If a score changed between two submissions and the crop, angle and light all look the same on review, distance is the first thing worth checking, because it is the variable most likely to have drifted without anyone noticing they had drifted. The tell is proportion: does the subject fill noticeably more or less of the frame in one photo than the other, relative to a fixed reference point in the background? If it does, the two photos were not taken from the same spot, and the honest read of a score difference is "position moved," not "something changed." This is a cheap check to run before concluding anything else about a result.

The practical version

Pick a distance and write it down - a number of steps, a mark on a wall, whatever survives being forgotten in three weeks. Use it every time you want a result you can compare against a previous one. Further back with a crop afterwards tends to be more forgiving than close in, because small errors in distance matter proportionally more the nearer the lens is. Rate Cock's own guide to taking an accurate photo walks through the mechanics of doing this for its own tool; the principle - fix the distance, then everything downstream of it means something - applies wherever the file ends up.