A bad headshot costs you clients before you say a word. A great one signals competence, approachability, and trust in under two seconds. The good news: you don't need a $500 studio session to get there. With the right prompt and a clean selfie as your base image, tools like Midjourney, Leonardo AI, Adobe Firefly, or DALL-E 3 can produce a polished, professional result you'd actually use on LinkedIn, your website, or a speaking bio.
The key is specificity. Vague prompts produce vague results. The prompts below are built to be direct — lighting, background, lens, mood, and output quality all called out explicitly so the model has no room to guess wrong.
The Prompt
Use this as your core headshot prompt. Paste it into your AI image tool of choice. If the tool supports image uploads (like Midjourney with --cref or Adobe Firefly's "Generative Match"), attach your source photo to anchor your likeness.
Professional headshot portrait of a [man/woman/person] in their [30s/40s/etc.], wearing a [navy blazer / white button-down / smart casual top], photographed against a [soft gradient gray / clean white / blurred office interior] background. Studio lighting with a soft key light from the left and a subtle fill light, creating gentle shadows that define facial structure without harshness. Shot at eye level on a 85mm lens, shallow depth of field, sharp eyes, natural relaxed expression, confident posture, slight smile. Color graded to be warm and approachable. Photorealistic, ultra-sharp, professional LinkedIn-style headshot, 4K.How to Use It
- Start with a clean selfie. Face centered, decent natural light, no busy backgrounds. The AI uses this as a reference — garbage in, garbage out.
- Swap the bracketed fields for your specifics: age range, clothing color, background style. These three variables do most of the work.
- If your tool supports a reference image weight, start at 0.7–0.8. Higher values preserve your face more closely; lower values give the model more creative freedom.
- Run four to six generations before judging. Variation is normal. Look for the version where the eyes are sharp and the expression reads as natural.
- Check the hands and ears — these are where AI still stumbles. Crop tight to shoulders-and-above if you need to avoid artifacts.
- Export at the highest resolution available, then run it through Topaz Photo AI or Lightroom AI Denoise if you need to clean up any soft edges before using it.
Variations
Swap the core prompt's background, tone, or context to match where the headshot will live.
Outdoor editorial headshot, natural light, overcast sky, [person description], casual blazer over a plain tee, bokeh background of a city street or modern building exterior, 85mm, sharp eyes, relaxed confident expression, slightly desaturated warm tones, editorial magazine quality, photorealistic.Corporate executive headshot, dark charcoal suit, crisp white shirt, no tie, shot against a deep navy or dark gray seamless backdrop, dramatic Rembrandt lighting, strong jaw definition, authoritative but approachable expression, ultra-sharp, 4K, photorealistic.Creative professional headshot, relaxed smart-casual outfit in muted earth tones, shot in a bright airy studio with white walls and natural window light, soft shadows, genuine smile, eye contact with camera, film grain texture, warm color grade, lifestyle brand aesthetic, photorealistic.What to Watch For
- Eyes are the single most important element. If the eyes look glassy, slightly off-axis, or too symmetrical, regenerate. Viewers feel this immediately even if they can't name it.
- Lighting direction matters. "Soft key light from the left" consistently outperforms generic "studio lighting" because it gives the model a specific instruction to follow.
- Avoid requesting too many style descriptors at once. Pick a single tone — corporate, editorial, approachable — and commit to it in one prompt rather than mixing three aesthetics.
- If you're using a tool without image upload (text-to-image only), add specific physical descriptors: hair color, eye color, approximate age, skin tone. The more anchored the description, the more consistent the output.
- Run your final pick through a real person before publishing. Ask someone who doesn't know you if it looks credible. Fresh eyes catch what you've stopped seeing.
Tool recommendation: For the most consistent likeness, use a tool with face reference support — Midjourney (--cref flag), Leonardo AI (Image Guidance), or Adobe Firefly (Generative Match). Text-to-image-only tools are better for experimenting with style before you commit a reference photo.