LinkedIn portrait: how to keep the same person in a batch of AI images.

Answer first: use one approved reference photo, repeat the same identity anchors in every prompt, and lock the crop, light, lens feel, and background family before you make variations. Change one approved detail at a time. A batch looks like one person when the parts that identify them never move.

Why a batch can turn into several different people

AI can make each portrait look polished while quietly changing the features people use to recognize you. A new hairstyle, altered eye shape, smoother skin, a sharper jaw, or a different smile can be small in one image and obvious when the images sit together. The problem gets worse when every prompt asks for a fresh mood, location, outfit, and camera treatment.

For LinkedIn, consistency is more useful than variety. Start by deciding what cannot change, then make controlled versions for the places you may actually use them: a profile image, a team page, a speaker bio, or a simple announcement.

Build two cards before you generate

1. The identity card: details that must stay

Write this from an approved, current reference photo. Use facts you could point to, not compliments. Include face shape, eye spacing or other recognizable eye detail, nose and jaw shape, skin tone and natural texture, hairline and hair texture, glasses if you usually wear them, and your normal expression. The point is not to demand a flawless copy; it is to make a recognizable person the non-negotiable part of the request.

Example identity card: Preserve the reference person’s oval face, close-set brown eyes, straight nose, soft jawline, medium warm skin tone with natural texture, shoulder-length wavy dark hair, thin round glasses, and relaxed closed-mouth expression. Keep ordinary proportions and age cues.

2. The production card: details that should repeat

This is the house style for the set. Pick a chest-up crop, a simple camera-height viewpoint, soft daylight or even studio light, and one restrained background family such as pale gray, warm off-white, or a softly blurred real workspace. Choose ordinary work clothing that you truly wear. Keep the same crop and lighting direction until you have an approved baseline.

Example production card: Chest-up LinkedIn portrait, eye-level viewpoint, soft window light from camera left, calm pale-gray background, relaxed shoulders, simple navy shirt, natural skin texture, clear but unforced expression.

The reusable LinkedIn portrait prompt

Paste the identity card and production card below without rewriting them for each image. Replace only the bracketed approved variation.

Prompt: Create a realistic LinkedIn portrait from the approved reference photo. [Paste the identity card exactly.] [Paste the production card exactly.] This version may change only [one approved variation, such as a charcoal blazer, a warm off-white background, or a slightly wider chest-up crop]. The result should look current, recognizable, and plausible for a professional profile. Do not alter facial structure, eye shape, skin tone, hairline, hairstyle, glasses, age cues, or the person’s familiar expression. Do not add logos, job titles, awards, credentials, readable text, extra people, or props that make a claim.

Make the batch in this order

1. Approve a baseline. Generate a single neutral portrait using both cards. Do not make other versions until a person who knows you would recognize it quickly.
2. Save the exact wording. Keep the approved reference, identity card, production card, and baseline result together. This is your batch packet.
3. Vary one thing. For the next image, change only wardrobe, background shade, or crop. Do not change all three to chase a different mood.
4. Compare side by side. Put each candidate next to the baseline at the small size where LinkedIn will show it. Check the face before you judge styling.
5. Reject drift early. If the eyes, jaw, hairline, skin tone, glasses, or familiar expression have moved, return to the unchanged packet. Do not repair a drifting image by piling on more style instructions.

Three safe variations after the baseline works

Version A: profile image

Approved variation: a slightly tighter head-and-shoulders crop. Keep every word in both cards unchanged. This helps the face read in a small circular avatar without creating a different person.

Version B: team directory

Approved variation: the same chest-up crop with a warm off-white background. Use this only after the neutral background version is approved. The background can vary; the identity evidence and light direction cannot.

Version C: speaker bio

Approved variation: a modest charcoal blazer over the same ordinary top, with a slightly wider chest-up crop. Do not use a stage, microphone, book cover, event signage, or title unless it is true and you need it for the actual context.

The Faymas same-person check

Before you keep an image, ask these four questions in order:

1. Recognition: Would a colleague identify the person without reading their name?
2. Anchors: Do the eyes, face shape, jaw, hairline, skin tone, glasses, and normal expression still agree with the baseline?
3. Batch fit: Does the crop, light direction, and background feel like it belongs beside the baseline?
4. Truth: Does the image avoid invented workplace proof, labels, people, and claims?

If recognition or anchors fail, discard the version and regenerate from the unchanged packet. If batch fit fails, change only the production detail that drifted. If truth fails, remove the invented object or text rather than trying to make the claim look more convincing.

What to avoid

Do not build a set from unrelated selfies taken years apart. Do not ask each version for a new face, dramatic editorial lighting, a different haircut, or perfectly smooth skin. Do not treat a face-similar image as enough when it no longer has your familiar proportions or expression. And do not use fabricated employer marks, qualifications, client names, awards, or numbers to make the portrait seem more professional.

For a prompt that adapts a single portrait to your real work context, see professional headshot prompts. For a closer review of what changed in a generated image, use the identity-and-change audit.