What a professional headshot actually has to do

Most advice about profile pictures is about taste. Smile more, wear blue, find a brick wall. That advice is not wrong, it is just downstream of the part nobody says out loud: a headshot is a functional object with about three jobs, and if it does those it is fine, and if it does not, no amount of styling rescues it.

Job one: be recognisably you, in the room you are walking into

The photo exists so a stranger can match it to a person. That is the whole point. A recruiter has your profile open in one tab and your name in an inbox, and the picture is what joins them. A shot that flatters you into somebody else fails at the only thing it was for.

This is worth holding onto because it is the exact thing generative headshot tools trade away. They redraw your face from a model’s idea of a face like yours, and the result is frequently better looking and slightly not you. If you have ever seen a set of AI headshots where the jaw is subtly wrong, that is the mechanism.

The second half matters too. The same face needs different formality for a law firm partner page and a design studio’s about page. You are not picking “professional” in the abstract, you are picking the version of yourself that belongs in one specific room.

Job two: survive being shrunk

Here is the constraint that decides more than people expect. Your carefully composed portrait is almost never seen at the size you made it. It appears as a small circle next to a comment, in a list of search results, in a sidebar, in a meeting invite.

At that size, detail is gone. What is left is a silhouette and a value pattern: the shape of your head against the background, and where the dark and light areas fall. Which gives you two practical rules.

Keep the head large in the frame. If your head is a third of the frame height, it is a smudge in a 48 pixel circle. Something around 55 to 70 per cent of frame height is the range that reads at both sizes, and tighter is usually safer than looser.

Keep the background separated from you in value. A dark jacket against a dark wall merges into one blob when it is small. You do not need a bright background, you need a different one. This is the entire reason studio backdrops are mid grey: grey separates from both dark hair and light shirts.

Job three: get the geometry conventional

There is a reason professional headshots look alike. The conventions are not fashion, they are the accumulated answer to “where do people look first”.

Viewers go to the eyes, so the eyes want to be roughly 40 to 45 per cent of the way down the frame. Put them in the middle and the shot reads as a passport photo. Put them too high and the subject looks like they are falling out of the top.

Level the eye line. A head tilted two or three degrees reads as casual, which is sometimes what you want and usually not, and a tilt is one of those things that looks fine in the original and wrong the moment it sits in a grid next to other people’s photos.

Leave a little room above the head, but only a little. The classic amateur crop leaves a third of the frame empty above the hair. The classic overcorrection cuts the crown off. If your source photo has the top of the head cut off, the honest fix is a different photo, because the missing skull cannot be invented without making up a face.

For the record, the numbers our own tool frames to: the eye line at 42 per cent, the head filling 62 per cent of frame height for a standard crop, and the chin landing around 73 per cent. Those are not magic values, they are just the middle of the conventional range.

What actually makes the difference, ranked

If you are deciding where to spend effort, this is roughly the order that pays.

Light comes first, by a distance. Soft light from one clear direction does more for a face than any camera. A window on an overcast day is a large soft source and costs nothing.

Framing comes second, and it is free to fix later since it is only a crop.

Background comes third. Plain and separated in value beats interesting.

Clothing comes fourth, and mostly by not competing: no busy patterns, nothing that vanishes into the background.

Camera comes last. A recent phone in good light beats a good camera in bad light every time, and it is not close.

The parts you can fix afterwards, and the parts you cannot

Framing, background, colour balance, evenness of light across the face, shine on the forehead, a slightly crooked eye line: all fixable after the fact, because they are adjustments to information the photo already contains.

Focus, motion blur, a face turned away from the camera, a crown cropped out of frame, eyes closed: not fixable, because the information is not there. Anything that claims to fix them is generating a new face, with the identity trade that implies.

That split is why our tool takes the shape it does. It does not draw anything. It uses two small models to work out where things are, which pixels are background, hair, skin and clothes, and where 478 points on your face sit, and then does ordinary image processing inside those masks. Background replaced, key light and rim added, skin and teeth and hair corrected in their own masks, framed on the eye line, sharpened. About four seconds, on your own machine, and the person who comes out is the person who went in.

If you have a photo that gets the light and the focus right, drop it in and see what the rest looks like handled for you. If you do not have one yet, our guide to shooting a headshot with the phone you already have covers the two parts you cannot fix later.