Get a Better AI Headshot with the Right ChatGPT Prompt

Start with Your Real Photo as a Reference

Think of this as giving an artist a sketch. Without it, you're asking for a guess. Upload a clear, well-lit selfie to the AI tool. I mean a real photo, taken in front of a window or under a bright lamp. Not a grainy picture from your cousin's wedding ten years ago. The AI needs to see the actual structure of your face—your jawline, your eye shape, the distance between your nose and mouth. One client used a dark, shadowy photo and got back a headshot where her chin melted into her neck. We used a simple phone selfie against a plain wall, and the next result had perfect bone structure. That reference photo is your anchor.

Describe Your Face and Hair with Precision

Generic words create generic people. "Professional brown hair" could give you anything from a chestnut bob to a muddy, shapeless helmet. You must be specific. List three things: length, style, and color. Say "chin-length blunt bob, ash brown hair, with a deep side part." For your face, describe your most stable feature. Is your skin "porcelain with freckles across the nose and cheeks"? Are your eyes "deep-set, hazel, with pronounced crow's feet when smiling"? This isn't vanity; it's technical specification. I once prompted for "a friendly face" and got a terrifying, toothy grin. When I changed it to "a warm, closed-mouth smile that reaches the eyes," the person in the image looked approachable and genuine.

Specify Your Professional Setting and Lighting

The background and light define your entire vibe. "Office background" is useless. It could be a sterile cubicle or a mahogany law library. You need a scene. Try "modern co-working space with a slightly blurred plant in the foreground" or "neutral grey textured studio wall." Now, lighting. This is where most free AI headshots fail—they look flat and dead. You must command the light. Use photographer terms: "soft window light coming from camera-left, creating a gentle catchlight in the eyes" or "clean studio loop lighting with a soft shadow under the chin on the neck." This tells the AI to model your face with dimension. A headshot with directional light looks hired. A headshot with flat, overhead light looks like a mugshot.

Avoid These Two Common Prompt Mistakes

First, avoid contradictory commands. The AI gets confused and averages them out into a mess. "A serious, authoritative expression but also warm and giggling" will give you a strained, uncanny valley smirk. Pick one primary mood. Second, stop using vague adjectives for quality. Words like "high-quality," "amazing," or "photorealistic" are wasted tokens. The AI is already trying to make a photo. Instead, use concrete technical nouns that affect the image construction. Swap "high-quality" for "shot on a 85mm lens" to get pleasing facial proportions and background compression. Replace "professional" with "Rembrandt lighting pattern" for a classic, dramatic shadow. The machine understands lenses and light patterns. It doesn't understand marketing fluff.

• Mistake 1: "Friendly and serious CEO" → Fix: "Approachable CEO with a confident, closed-mouth smile."
• Mistake 2: "Ultra-realistic photo" → Fix: "Shallow depth of field, 100mm portrait lens, sharp focus on the eyes."

Combine the Elements into Your Final Prompt

Now, build your recipe. Structure it like a photographer's brief. Lead with the subject, then the scene, then the technical specs. Here’s a template you can copy and fill in: "A professional headshot of a [man/woman/person] with [specific face/hair description]. They are in a [specific setting] with [specific lighting description]. Shot on a [lens type like 85mm or 100mm lens], shallow depth of field, sharp focus on the eyes, [specific expression]." Let's make one real. "A professional headshot of a woman with a long oval face, high cheekbones, and long curly black hair tied in a low bun. She is in a bright modern art gallery with white walls, soft diffuse light from a large overhead source. Shot on an 85mm lens, shallow depth of field, sharp focus on her green eyes, expressing calm intelligence with a slight smile." See the difference? That prompt can be executed.

The right prompt doesn't just request a picture—it directs a virtual photoshoot. You've chosen the lens, set up the lights, and styled the subject. Your takeaway is this: be the art director, not just the client. Start with a good photo, describe what you see in the mirror with ruthless detail, design the light and space around you, and speak the camera's language. Then hit generate. The result will look like you, on your best day, ready for work.