What a photograph hides in a lookalike comparison: distance, light, expression

Here is the central problem with comparing two photographs of two faces: a photograph is not a record of a face. It is a record of a face, a lens, a distance, a light source, an expression, a moment, and a chain of processing decisions, all fused into one image with no way to separate them afterwards.

Two pictures of the same person can look like two people. Two people can be photographed into looking like one. Neither outcome requires anyone to be doing anything dishonest. It falls out of how cameras work.

What to take away

Camera distance changes the shape of a face

This is the single largest effect and the one most people have never had explained to them.

A face is three-dimensional, and the apparent proportions of a solid object change with how far away the viewer stands. The nose sits several centimeters closer to the camera than the ears.

When the camera is close, that gap is large relative to the total distance, so near features project bigger and far ones smaller. The nose enlarges, the ears recede, the face narrows toward the edges, and the head looks rounded and forward-leaning.

Move the camera back and use a longer lens to fill the frame the same way, and the relative depth difference shrinks toward nothing. Features flatten into a single plane. The face widens, the nose settles, the ears come forward, the jaw reads straighter.

The same skull produces both images; nothing about the person changed, only how far away the camera stood.

A selfie at arm's length and a portrait from across a room are not comparable evidence about anyone's face. Most casual comparisons put exactly those two things side by side.

Productions exploit the same effect deliberately when they need a double to read as the lead.

What else the image is carrying

Any one of these can move a face further than the difference between two actual people. In a typical side-by-side, several of them differ at once and none of them are stated. A publicity image adds a further designed layer on top before any of this begins.

Why the same person can look like a stranger

Ask anyone about their passport photograph. The complaint is universal, not vanity. Identity documents are made under conditions deliberately stripped of everything that carries personality: neutral expression, flat frontal light, fixed distance, no head tilt, often no smile.

What remains is a face with its habitual configuration removed, and that configuration is a large part of what people recognize you by.

The general principle: familiar face recognition draws on many images across many conditions, and it tolerates variation well. Why an unfamiliar population makes this harder again is a separate and compounding problem.

Unfamiliar face recognition or matching has none of that redundancy: one image, one set of conditions, and those conditions are inseparable from the person.

How side-by-side layouts manufacture a match

The comparison format is not neutral. Assembling one involves a series of choices, each of which can be made to increase apparent similarity:

None of that requires bad faith. Doing the ordinary work of making a tidy graphic produces most of it automatically.

What a fair comparison would need

If someone truly wants to compare two faces, not two photographs, they must hold conditions constant.

The camera distance and focal length, lighting, neutral expression, and head angle must match. Processing must match, retouching must be absent, and they need multiple images of each person, not one.

That is a controlled photographic session, not a screenshot. But the result still would not tell you everything.

Even a perfectly matched pair of images supports only one claim: how two faces photograph under one condition. It does not show that two people look alike in life, in motion, or to anyone else. No measurement resolves resemblance: no unit, no threshold, no test that comes back positive.

Resemblance is not identity

The distinction runs through everything on this site and it matters most here, because the comparison format is the exact place where it collapses.

Two images side by side invite a binary reading: same or not same. The honest output of a photographic comparison is neither. It is a statement about two images under unknown conditions.

Treating that as a verdict on whether two people are the same person turns casual face-matching into wrongful accusation. The accuser usually never learns whether they were wrong.

For this reason a comparison should never be used to identify a private individual from a photograph, and this site does not describe methods, tools or services for doing so. That holds regardless of the reason given.

The practical version

If you are going to compare, do it honestly:

The presentation-layer version of these problems, how a published pairing is built and why it persuades, is covered in celebrity lookalikes comparison.

Common questions

Which single factor moves a face the most?

Camera distance, in most cases. It is also the factor almost never recorded, which is why so few circulating comparisons can be assessed at all.

Is a passport photograph a good comparison image?

It is consistent, which helps, and it strips out the habitual expression and posture that people are actually recognized by. Consistent is not the same as informative.

Can editing software undo any of this?

No. The lens, the light and the expression are fused into the pixels, and nothing in the file records what they were.

What is the strongest test available to an ordinary reader?

Video of both people, several seconds each. Motion carries individual information that photographs discard, and most still-image resemblances do not survive it.