AI Headshot Prompts for LinkedIn That Do Not Look AI

Most "AI headshot" results you have seen are instantly identifiable, and it is almost never the face that gives them away. It is the background — a fake-blurred office that has no actual geometry behind it — plus lighting that arrives from everywhere at once, a suit with no fabric weave, and skin rendered like a wax casting. A recruiter does not consciously diagnose any of that. They just get a flicker of "something is off" and move on.

That flicker is the entire problem, because a LinkedIn photo has exactly one job: to make a stranger trust that you are a real, competent person before they read a single word. A photo that reads as synthetic does the opposite of that job, no matter how flattering it is.

The good news is that a professional headshot is one of the easiest things to generate convincingly, because a real headshot is a highly constrained, well-documented lighting setup. You are not asking the model to invent a scene. You are asking it to reproduce a studio configuration that exists in millions of its training images. You just have to name that configuration instead of asking for "a professional headshot."

Why "Professional Headshot" Is a Useless Prompt

Type "professional headshot of me, business attire" into any image model and you get the average of every stock corporate photo ever taken: even light, gray gradient background, glassy skin, a smile with no muscle behind it. It looks professional the way a mannequin looks professional.

The word "professional" carries no visual information. It does not tell the model where the light is, how far away the camera is, what lens is on it, or what the background actually is. Every one of those unspecified variables gets filled in with the model's blandest average — and blandness is exactly what reads as fake.

A usable headshot prompt names five things explicitly:

1. The lighting setup — not "good lighting," but the actual arrangement.
2. The lens and distance — headshots have a specific look because they are shot with a specific focal length.
3. The background as a real place — with depth, not a blur filter.
4. The wardrobe as fabric — texture, weave, fit, not just "a suit."
5. Skin as skin — pores, minor asymmetry, no retouching.

Miss any of the five and the model defaults, and every default pushes toward the stock-photo average.

The Lighting Language That Does the Most Work

Real headshot photographers use a handful of named setups. Naming one of them in your prompt is the single highest-leverage change you can make, because it collapses hundreds of ambiguous variables into one recognizable configuration.

Rembrandt lighting — key light high and about 45 degrees off-axis, producing a small triangle of light on the shadowed cheek. This is the classic serious-executive look. Say it by name: "Rembrandt lighting, key light high and 45 degrees camera-left, small catchlight triangle on the shadow-side cheek."

Loop lighting — key light slightly off-center and just above eye level, casting a small nose shadow that loops down toward the corner of the mouth. Softer and friendlier than Rembrandt; the default for approachable corporate portraits.

Clamshell lighting — one soft source above and one reflector below, filling under-eye shadows. Flattering, modern, slightly beauty-campaign. Good for consultants and client-facing roles, less good if you want to look like you have opinions.

Window light — a single large soft source from one side, deep natural falloff on the other. The most natural-looking of the four and the one that reads least like a photo studio, which is often what you want in 2026.

Whichever you pick, add the falloff. Light that hits a face has to go somewhere, and describing where it stops is what makes it look physical: "soft falloff into the shadow side, background about two stops darker than the face."

The Lens Is Not a Detail

Headshots look like headshots partly because of compression. An 85mm or 105mm lens at portrait distance flattens facial features slightly and separates the subject from the background in a way a phone camera at arm's length does not. Wide lenses close to the face enlarge the nose and pull the ears back — the reason selfies rarely look like you.

So specify it: "shot on an 85mm lens at f/2.8, subject three feet from camera, shallow depth of field with the background falling into soft, optically continuous bokeh."

The phrase "optically continuous" is doing real work there. The dead giveaway of a fake headshot is a background that is uniformly blurred like a Photoshop gaussian, rather than blurred progressively with distance the way an actual lens renders it. Asking for depth-dependent blur pushes the model toward the latter.

Give the Background a Real Location

"Blurred office background" produces a smear. Name a place with actual objects in it, and let the depth of field do the blurring:

- "background: a modern open-plan office twenty feet behind, glass partition walls, a row of pendant lights, one green plant, all thrown well out of focus"
- "background: a north-facing window with an overcast city skyline behind it, softly defocused"
- "background: a plain warm-gray studio backdrop with a subtle light gradient falling off toward the lower corners"

The third one is the safest choice for most roles, and note that even the "plain" version specifies a gradient and a direction. A perfectly uniform backdrop is a rendering artifact; real seamless paper always has falloff.

Wardrobe as Fabric, Not as Category

"Wearing a suit" gets you a suit-shaped region of flat color. Fabric has weave, drape, and the specific way it catches light:

- "charcoal wool suit jacket with visible worsted weave, natural shoulder, slight break at the lapel roll"
- "navy merino crewneck, fine knit texture visible at the shoulder seam"
- "white cotton oxford shirt, subtle texture in the placket, top button open, faint natural creasing at the elbow"

That last detail — creasing — matters more than it sounds. Clothing that has never been sat down in is one of the most reliable tells of a generated image. A small amount of natural wrinkle is what tells the eye a body is inside the clothes.

Skin, Because This Is Where It Falls Apart

Every image model biases toward smooth, even, symmetrical skin, because a huge share of its training data is retouched. Left unspecified, you get that bias every time. The fix is the same vocabulary that solves it everywhere else — described in detail in how to fix the plastic skin problem — and for a headshot it is doubly important, because the face is 60 percent of the frame with nowhere to hide.

Add: "visible pore texture across the nose and cheeks, slight asymmetry in the eyes and jawline, fine lines at the outer eye, natural skin tone variation, no beauty retouching, no airbrushing."

Then add the negative: "not CGI, not a 3D render, no plastic skin, no perfect symmetry, no digital smoothing."

Four Prompts You Can Use Directly

The executive. "Photorealistic corporate headshot of the person in the reference image. Charcoal wool suit jacket with visible worsted weave over a white cotton oxford shirt, top button open. Rembrandt lighting, key light high and 45 degrees camera-left, small catchlight triangle on the shadow-side cheek, soft falloff into shadow. Background: warm-gray seamless studio backdrop about two stops darker than the face, subtle gradient falling toward the lower corners. Shot on an 85mm lens at f/2.8, subject three feet from camera. Visible pore texture, slight facial asymmetry, fine lines at the outer eye, no retouching. Composed as a 4:5 vertical crop, head and upper chest, eyes on the upper third line. Neutral confident expression, mouth closed, direct eye contact. No CGI, no plastic skin, no digital smoothing."

The approachable operator. Same technical block, but swap in: "Loop lighting, key light slightly camera-right and just above eye level, small nose shadow looping toward the corner of the mouth. Navy merino crewneck, fine knit texture. Background: modern open-plan office twenty feet behind, glass partitions and pendant lights thrown well out of focus. Genuine closed-mouth smile with visible muscle engagement around the eyes."

The natural-light founder. "Window light from camera-left, single large soft source, deep natural falloff on the camera-right side of the face. Background: a north-facing window with an overcast city skyline, softly defocused. Light gray marl sweater over a collared shirt. Shot on a 50mm lens at f/2.0, slightly closer framing." This one reads least like a studio and best on a personal site.

The environmental portrait. Wider crop, more context: "Three-quarter body, seated on the edge of a desk, hands relaxed. 50mm lens at f/2.8, six feet from camera, workspace visible and defocused behind. Practical light from a desk lamp camera-right plus ambient window fill." Use this as the second image on a profile, not the primary.

Framing and Crop, Since LinkedIn Will Crop You Anyway

LinkedIn displays the profile photo as a circle. Anything you place near the corners of a square image is going to be cut. Generate at 4:5 or 1:1, keep the head centered horizontally, put the eyes on the upper-third line, and leave a comfortable margin above the head. If your generated image has your shoulders clipped at the frame edge, it will look cramped once the circular mask lands on it.

For resolution, generate larger than you need and let the platform downscale — a downscaled image keeps its texture, an upscaled one invents it. The full breakdown of what to generate at for which platform is in the aspect ratio and resolution guide.

One Thing to Get Right Before You Publish It

A generated headshot is a photo of a person who does not exist unless you anchor it to a reference image of yourself. If you are using this on a professional profile, that anchoring is not optional — it is the difference between an enhanced portrait and a fabrication. Upload a clean, well-lit reference photo of your actual face, and keep the same locked description clause across every generation in the set so all your professional images match. The technique for that is in getting consistent AI photos across generations.

There is also a disclosure question here that is worth two minutes of your time before you upload anything to a professional network. It is covered in do you have to label AI photos.

Where PROMPTMVSTR Fits In

The lighting-and-lens discipline above is the same structure every prompt in the archive is built on — a named setup, a real lens, a background with depth, and skin described as skin. If you want the professional set and the lifestyle set to look like the same person shot on the same day, the Virtual Photoshoot carries one reference upload across a whole batch, and the designer fashion prompts cover the wardrobe side in far more depth than a single headshot needs.