A professional headshot has three tells that signal “real photo”: a clean background, soft three-point or window lighting, and an 85mm-ish lens that compresses the face naturally. Wide lenses warp jawlines, busy backgrounds add AI clutter, and overhead key light cooks the skin. The 12 prompts below force all three for different use cases — LinkedIn, speaker bios, press kits, team pages — so the output reads as photography. For same-person variant sets across uses, see the AI realistic portrait workflow.
TL;DR
- Always pin lens + aperture (
85mm f/2.0), background type, and lighting in the prompt. Defaults drift toward dramatic wide lenses and over-smoothed skin. - For skin realism in 2026, Flux 2 leads on face-forward portraits; Nano Banana Pro (Google Gemini 3 Pro Image) is the sharpest on micro-detail; Midjourney V8.1 is fastest and renders native 2K.
- To make a stranger look like you, a prompt is not enough — use a likeness-trained tool (Aragon, BetterPic, PhotoPacks) that learns your face from selfies, or Midjourney Omni-Reference (
--oref). - Add
natural skin texture, visible pore detailto every prompt to kill the plastic AI look. - For darker skin tones, default lighting over-brightens; add
lighting calibrated for darker skin tones (lower key ratio, more fill).
Best for
- LinkedIn profile and resume headshots
- Speaker bios and conference programs
- Founder photos for press kits and About pages
- Internal team directories where consistency matters
Which model to use (June 2026)
These prompts are model-agnostic, but the engine changes the result. Current picks, as of June 2026:
| Goal | Best pick | Why |
|---|---|---|
| Maximum skin realism, face-forward | Flux 2 (Flux 1.1 Pro Ultra for 4MP) | Leads on pore-level skin texture; Raw mode favors natural skin |
| Sharpest micro-detail + clean text | Nano Banana Pro (Gemini 3 Pro Image) | Stock-photo-grade detail on portraits |
| Fastest, native 2K, easy iteration | Midjourney V8.1 | Renders ~4-5x faster than V7; 2K HD without upscaling |
| Looks exactly like you from selfies | Aragon / BetterPic / PhotoPacks | Trains a model on your face; ~$30-50 for 40+ shots |
| Quick draft inside a chat | ChatGPT (GPT Image) / Gemini | Convenient, but small face distortions still show at full size |
Midjourney V8.1 shipped April 30, 2026 with native 2K HD output and roughly 4-5x faster renders than V7; plans run Basic $10, Standard $30, Pro $60, Mega $120/month (20% off annual). General models like these cannot reproduce your exact face from a text prompt alone — that is what the likeness-trained tools or Omni-Reference are for (see Same-face consistency below).
1. Corporate standard
professional headshot of {subject}, neutral light grey backdrop, soft three-point studio lighting, 85mm f/2.8, sharp focus on eyes, business attire (navy blazer, white shirt), confident slight smile, natural skin texture with visible pore detail, photo realistic, 4:52. Tech / smart casual
professional headshot of {subject}, soft natural window light from camera left, blurred warm office background, 85mm f/2.0, smart casual top (knit polo or merino crewneck), warm friendly expression, sharp focus on eyes, natural skin, candid photo realism, 4:53. Editorial magazine
editorial headshot of {subject}, dramatic single side light with subtle fill, dark grey seamless backdrop, 85mm f/1.8, intense direct gaze into camera, slight head tilt, magazine portrait style, natural skin texture, 4:54. Founder press kit
founder portrait of {subject}, leaning against a desk in a softly lit office, large window background out of focus, 50mm f/2.0, business casual (open collar shirt), thinking expression looking off-camera, environmental portrait, photo realism, 3:25. Speaker bio (conference)
speaker portrait of {subject}, mid-presentation feel, soft stage spotlight from above front, dark blurred audience background, 85mm f/2.0, animated speaking expression, hand gesture visible, photo realism, 4:56. Team page (consistent series)
team headshot of {subject}, same studio setup for series consistency: light grey backdrop, soft butterfly lighting at +30 degrees, 85mm f/2.8, neutral expression with slight smile, business casual, sharp eyes, natural skin texture, photo realism, square crop7. Outdoor natural light
outdoor headshot of {subject}, soft overcast daylight (no direct sun), urban park background blurred out, 85mm f/2.0, jacket and open collar shirt, relaxed natural expression, sharp focus on eyes, photo realism, 4:58. Black-and-white classic
black and white headshot of {subject}, single soft north-facing window light, deep grey backdrop, 85mm f/2.0, classic three-quarter pose, contemplative expression, fine grain, fine art portrait style, 4:59. Creative / artist portrait
creative portrait of {subject}, moody warm tungsten light, dark brick wall background, 50mm f/1.8, casual layered outfit (t-shirt + jacket), unposed expression mid-thought, slight cinematic color grade, photo realism, 4:510. Academic / faculty
academic faculty headshot of {subject}, soft window light from camera left, blurred bookshelf background, 85mm f/2.8, blazer over button-up, calm authoritative expression, sharp eyes, natural skin texture, photo realism, 4:511. Healthcare / medical bio
healthcare professional headshot of {subject}, neutral white backdrop, soft even clinical lighting, 85mm f/2.8, white coat over scrubs or shirt, warm trustworthy expression, sharp eyes, no harsh shadows, photo realism, 4:512. Founder of color, natural skin tone
headshot of {subject, e.g., Black founder / South Asian founder}, neutral mid-grey backdrop, soft three-point lighting calibrated for darker skin tones (lower key ratio, more fill), 85mm f/2.0, business casual, warm confident expression, accurate skin tone, natural texture with pore detail, photo realism, 4:5How to refine
Specify the lens (85mm, f/2.0) and the background type (seamless, blurred warm, dark backdrop) explicitly — diffusion models drift toward dramatic wide lenses by default. Ask for natural skin texture, slight pore detail to fight the airbrushed AI look; on Flux, Raw mode reinforces this further. For non-white subjects, add lighting calibrated for [skin tone] and accurate skin tone — default model settings often over-brighten and lose detail.
Same-face across shots
For a same-person variant set across LinkedIn, speaker bio, and team page, three approaches work in 2026:
- Midjourney Omni-Reference — add
--oref [image URL] --ow 100on V7/V8 to lock the face; dial--owdown to ~70 when you want outfit or pose freedom. Note: Omni-Reference costs about 2x the GPU time of a normal render, and the result is not compatible with Vary Region, Pan, or Zoom Out. - Likeness-trained tools (Aragon, BetterPic, PhotoPacks) — upload 10-20 selfies, get 40+ on-model shots; best when the face must be you, not a lookalike.
- Seed + identity anchor — on seed-based engines, lock the seed and reuse the same identity-describing sentence in every prompt for a near-consistent look.
Common mistakes
- Background too busy — competes with the face and reads “AI”
- Wrong lens (35mm or wide) distorts jaw and nose
- Plastic, over-smoothed skin from default model settings; ask for texture explicitly
- Smile too rehearsed; “slight” or “subtle” smile keywords help
- Lighting not calibrated for darker skin tones — defaults wash out detail
FAQ
Can a text prompt alone make the AI look exactly like me?
No. Text-to-image models (Midjourney, Flux, Nano Banana, ChatGPT) generate a plausible person, not your specific face. To get your likeness you need a likeness-trained tool that learns from your selfies (Aragon, BetterPic, PhotoPacks, ~$30-50 for a batch) or an image reference such as Midjourney Omni-Reference (--oref).
Which model gives the most realistic skin in 2026? For face-forward close inspection, Flux 2 leads on natural skin texture (use Raw mode). Nano Banana Pro (Google Gemini 3 Pro Image) renders the sharpest micro-detail. Midjourney V8.1 is the fastest and outputs native 2K HD without an upscale step.
Why does my AI headshot still look fake?
Usually one of three things: an over-smoothed default skin pass (add natural skin texture, visible pore detail), a wide lens that distorts features (force 85mm), or a cluttered background (force seamless or blurred). The prompts above bake all three fixes in.
Are AI headshots OK to use on LinkedIn? LinkedIn allows AI-assisted profile photos as long as they represent you accurately. Likeness-trained tools that learn from your real selfies are the safest route for a profile picture; a generic prompt-only face is fine for stock or placeholder use but is not “you.”
What aspect ratio should I export?
Use 4:5 for LinkedIn and most social profiles (it crops cleanly to a circle), square for team-directory grids, and 3:2 for wider press-kit or environmental shots.
Related
- Realistic portrait prompts — broader portrait library
- Realistic portrait lighting prompts — lighting recipes for portraits
- Business male portrait prompts — corporate male variant
- AI realistic portrait workflow — production tutorial
- Environmental Portrait Prompts Where Background Tells Story
- Street Portrait Prompts: 10 Candid Documentary Templates
Tags: #Portrait #Realistic