Verified against Nano Banana / Gemini 3.1 Flash Image · 2026-08-10
Turn a casual selfie into a professional headshot without changing the person's actual face
An identity-preserving edit brief for an uploaded selfie — swaps background, attire, and lighting to a studio headshot look while explicitly locking the person's real facial identity in place.
The prompt
Ready to copy — highlighted parts are example details you can swap.
You are editing the selfie I've uploaded in this conversation into a professional headshot suitable for a LinkedIn profile photo and company website team page. This is an edit of the same person, not a new person who merely resembles them — their exact facial structure, proportions, skin tone, and any distinguishing features (moles, scars, asymmetry) must be preserved in identity, even as everything around them changes. SOURCE PHOTO a phone selfie taken indoors under mixed overhead light, casual t-shirt, slight upward camera angle WHAT CHANGES Replace the background with a softly blurred neutral charcoal-grey studio backdrop, and change their clothing to a dark navy blazer over a plain white collared shirt. Adjust the lighting to read as a professional headshot session — soft, even, flattering key light with gentle falloff, not the harsh or uneven light a phone camera in a casual setting typically produces. WHAT DOES NOT CHANGE Their face shape, facial features, expression style, skin texture, hair (unless told otherwise), and apparent age must stay exactly as they are in the source photo. Do not slim, smooth, symmetrize, or idealize their face beyond what light skin evenness only — keep visible pores, natural under-eye area, and any freckles exactly as they are allows — a professional headshot should look like a great photo of this specific person on their best day, not a different, more generic-looking person. FRAMING Standard headshot crop — head and upper shoulders, eyes roughly on the upper third line, looking toward camera or at a natural slight angle matching the source photo's head angle if it was already close to camera-facing. IF SOMETHING CAN'T BE DONE CLEANLY If the source photo's angle, lighting, or resolution makes it genuinely impossible to produce a clean studio-quality result without materially altering their face to compensate, say so directly rather than delivering a subtly different-looking person and calling it done. OUTPUT One headshot-cropped image of the same person, professionally lit and dressed, identity fully intact.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Nano Banana's conversational edit mode works from an uploaded reference image, but a large simultaneous change to background, lighting, and clothing gives the model more room than a small local edit does, and more room is exactly where identity drift creeps in — under a big enough scene change, an image model will sometimes quietly regenerate a face that's subtly more symmetrical, younger, or generically attractive rather than a strict edit of the one that was uploaded, because that idealized direction is what its training data skews toward for "professional headshot" as a category. Stating explicitly that this must remain the same person, not a person who merely resembles them, and naming the specific features that must stay put (moles, asymmetry, skin texture) gives the model a concrete checklist to hold onto instead of an implicit expectation it can drift away from unnoticed. The retouching-limits field matters for the same reason a photographer's retouching brief matters with a human editor: "professional headshot" alone is ambiguous between a light touch-up and a heavily smoothed beauty-retouch look, and without specifying which one is wanted the model defaults toward the more aggressively polished interpretation it's seen more often labeled that way. Finally, giving the model explicit permission to say the result can't be produced cleanly — rather than forcing a silent best-effort attempt — matters because a headshot with a subtly wrong face is worse than no output at all for this specific use case, where the entire point is that the viewer recognizes the actual person.
Verified against
Nano Banana / Gemini 3.1 Flash Image Gemini 3.1 Flash Image · 2026-08-10
Changelog
- 2026-08-10 — Initial publish, verified against Nano Banana (Gemini 3.1 Flash Image) on a casual-selfie-to-headshot edit.
Need this built into your business?
If a prompt isn't enough — what Scult builds, built and maintained for you — that's Scult's day job.
EXPLORE WHAT SCULT BUILDS
