Cyberpunk Sunglasses Portrait: The Exact AI Prompt Revealed

Extreme close-up portrait of a woman with pale skin, wearing round metallic-framed sunglasses with reflective yellow-green gradient lenses showing digital grid reflections and visible green eye with dramatic black eyelashes behind the left lens, glossy dark navy blue lips slightly parted showing teeth, thick dark eyebrows, short black choppy bangs, wearing a black choker with white text "GRUAT", cool blue cinematic lighting from the left side creating soft shadows on the right side of face, shallow depth of field, hyper-realistic skin texture with visible pores and subtle freckles, cyberpunk aesthetic, fashion photography style, moody atmosphere, high contrast --ar 9:16 --style raw --s 250

How I Burned Through 23 Attempts to Get This Right

Last Tuesday at 11:47 PM, my client Elena from that Berlin synthwave label messaged me. "We need the cover. Tonight." She'd seen a reference on Pinterest. Something about reflective sunglasses, blue mood, that whole cyberpunk aesthetic I'd been playing with.

I was pretty confident. Too confident.

First attempt? Garbage. The sunglasses looked like plastic toy glasses from a gas station. Second through eighth? Lenses were wrong color, wrong shape, wrong everything. By attempt fifteen I was mainlining espresso and questioning my career choices.

The problem wasn't the prompt length. It was the *specificity*.

See, AI image generators don't understand "cool sunglasses." They need "round metallic-framed sunglasses with reflective yellow-green gradient lenses showing digital grid reflections." Every. Single. Detail.

Why This Prompt Actually Works (I'm Not 100% Sure Why, Honestly)

Here's the thing about reflective surfaces in AI generation. They're *really* hard to control. The model wants to either make them mirror-perfect (boring) or completely opaque (useless).

But when you specify "visible green eye with dramatic black eyelashes behind the left lens," something clicks. The system understands layers. Reflection *and* transparency. It's basically telling the AI: "This is glass, not a wall."

The blue lighting direction matters more than I expected. "Cool blue cinematic lighting from the left side" creates that cyberpunk mood without needing neon signs or rain. (Though rain is cool. Street portraits with weather effects are a whole other rabbit hole.)

That choker with text? Total accident in my final working version. The prompt said "black choker with white text" and somehow "GRUAT" appeared. I kept it. Looks intentional, right?

Anyway.

The Technical Breakdown Nobody Asked For

Skin texture was driving me crazy. Early versions had that weird smooth porcelain look. Like mannequins. Adding "hyper-realistic skin texture with visible pores and subtle freckles" fixed it immediately.

Don't quote me on this, but I think the pore detail triggers the model to use higher frequency details across the whole image. Everything gets sharper.

The lip color specifically needs to be "glossy dark navy blue" not just "black lips." Black goes flat. Navy blue catches the blue environmental light and creates dimension. Small difference, huge visual impact.

(Side note: why does "glossy" work better than "shiny"? No idea. But it does. I tested both. Twelve times.)

And the aspect ratio—9:16 is non-negotiable for this composition. Vertical format forces the extreme close-up crop that makes the sunglasses dominate. Square format wastes space. Landscape format makes it look like a weird album cover from 2003.

How to Customize This Prompt for Your Project

So you've got the base prompt. Now what?

Change the lens color: "reflective magenta-orange gradient" gives completely different energy. More sunset vibes. Less corporate dystopia.

Swap the lighting: "warm amber lighting from below" creates horror movie tension. "Harsh overhead fluorescent" goes full office-drone-gone-rogue.

The choker text is actually controllable if you're patient. "Black choker with white text 'RESIST'" or "black choker with white text 'VOID'"—just keep regenerating. Took me 4 tries to get readable text consistently.

Want to push further into futuristic fashion territory? Add "metallic facial piercings" or "holographic makeup accents." I tried "circuit board patterns subtly visible on skin" once. Looked amazing. Took 8 attempts.

Professional Applications That Actually Pay

Elena used this for her label's EP cover. But I've since sold variations for:

—Beauty brand social campaigns (swapped sunglasses for futuristic eyewear)
—Tech startup hero images (added holographic UI elements in reflections)
—Fashion editorial spreads (changed lip color to metallic copper)
—Music festival posters (rainbow lens gradients, because obviously)

The key is treating this as a *system* not a one-off. Same base structure, different variables.

Pretty cool when it works.

One thing I learned the hard way: clients always want the sunglasses to show *their* logo in the reflection. AI can't do that reliably yet. I tried Midjourney, DALL-E 3, even Leonardo.ai for inpainting. Nope. Photoshop still required for that specific detail.

Honestly? That's fine. The AI gets you 90% there. The last 10% is where you justify your rate.

Final Thoughts From Someone Who Should Sleep More

This prompt took me from "I guess AI art is broken" to "okay this is actually useful" in about three hours of iteration. The difference was stopping halfway through, getting coffee, and actually *describing what I saw* in the reference image instead of what I *thought* the reference represented.

Specific beats poetic. Every time.

Try it. Break it. Make it yours. And if you get better results than me, don't tell Elena. I need that retainer.

Questions? Drop them below. I check comments between espresso shots.

Seriously.