Your Two Eye Colors, Perfectly Captured in AI Headshots

Upload Clear, Well-Lit Source Photos

Think of your source photos as the raw ingredients for a recipe. Garbage in, garbage out. You need crisp, clear pictures where your eyes are visible and well-lit. A blurry selfie taken in a dim bar won't cut it. The AI needs to see the subtle details—the starburst pattern in your left iris, the darker ring around the edge of your right one. I tell my in-person clients to aim for the quality of a passport photo, but with better lighting and a neutral expression.

How many photos should you upload? More is better, but only if they're good. Three to five excellent shots from different angles beats ten mediocre ones. Get one straight on, one with your head slightly turned, maybe one with a gentle smile. Use window light or stand under a bright, neutral ceiling light. I once had a client whose reference photo showed one eye in shadow; the AI gave him one bright blue eye and one that looked murky gray. We had to start over. The lesson? Light both eyes equally.

Use the Description Box to Specify Eyes

This is your control panel. Don't leave it blank. Be specific, plain, and direct. "Brown hair, blue eyes" is a start, but it's not enough for heterochromia. You must spell it out. The AI reads text literally, so give it a clear, unambiguous command.

What's the best way to phrase it? Try this: "The subject has complete heterochromia. The left eye is bright green. The right eye is dark brown." Or, "One eye is hazel with gold flecks, the other eye is steel blue." Describe what you see. If your eyes are sectoral heterochromia (two colors in one iris), say so: "My right eye is mostly blue with a large wedge of amber at the 2 o'clock position." This detail gives the AI a map to follow.

Review the First Drafts for Color Accuracy

When your first batch of images arrives, ignore the hair, the clothes, the background for a moment. Zoom in. Look only at the eyes. Are the colors distinct? Are they in the correct sockets? I've seen results where the colors were swapped—the green eye ended up on the wrong side of the face. It happens. Check for a weird, uniform color wash where both eyes are an average, muddy mix of your two real shades. That's a common first-draft fail.

Here's a trick: pull up your best source photo on your phone. Hold it next to the AI-generated headshot on your monitor. Compare them directly, eye to eye. Is the warmth the same? Is the saturation accurate? One of my studio clients has stunning central heterochromia—a ring of gold around her pupil inside a sea of green. The first AI drafts made her eyes a flat, solid lime. We needed to specify "green eyes with a distinct gold ring around the pupil" in the next round to get it right.

Avoid the 'Same-Color' Editing Mistake

You get your drafts back and the eyes are wrong. The instinct is to use the platform's editing tools to "fix" them. You might try to color-correct one eye, then use the same adjustment on the other. Stop. This is how you create the "uncanny valley" effect, where both eyes become an identical, unnatural hue. You are editing two unique features, not one.

Treat each eye as a separate edit. If the green eye is too dull, increase its vibrancy slightly. If the brown eye looks too red, adjust its temperature towards yellow. Make micro-adjustments independently. I watched a graphic designer do this on her own headshot; she used a single brush to "even out" her eye color, and the result looked flat and fake, like doll's eyes. The beauty is in the difference. Your two colors tell your story—don't erase that.

Get a Gallery That Reflects the Real You

The goal isn't one perfect shot. It's a collection of usable, authentic-looking portraits where your unique eyes are consistently correct. Aim for a gallery of 5-8 final images. You want variety in expression and angle, but unwavering accuracy in this one crucial detail. A slight smile, a serious look, a three-quarter turn—in all of them, your eyes should be unmistakably yours.

Why does a gallery matter? Because you need options. One shot might work for LinkedIn, another for a speaking bio, a third for a personal website. They all need to be recognizably you, and your eyes are the anchor of that recognition. When someone who knows you glances at the headshot, their thought should be, "Yes, that's them," not "Did they get new contacts?"

Getting this right is a process, not a one-click magic trick. It requires clear source material, precise instructions, careful review, and selective editing. But when you see that final gallery, where your distinct gaze looks back at you with all its real character, you'll know the effort was worth it. You get to keep a defining piece of yourself in every single shot.

Related, and worth reading next: Getting AI Headshots Right with Albinism: A Photographer's Guide.