Are You Still Writing 'Realistic Skin' in Your Cyberpunk

Extreme close-up portrait, cyberpunk female vocalist with rich deep brown skin showing natural pores and subtle sebum sheen, thick voluminous pink braided dreadlocks with individual strand separation and flyaway texture, magenta bandana with crisp white paisley print and "LOVE" typography wrapped tight across forehead obscuring eyes, angular chrome-framed cat-eye sunglasses with reflective pink mirrored lenses catching light, multiple silver spike hoop earrings stacked along ear cartilage with sharp geometric points, heavy oxidized silver chain necklaces with industrial spiked links and functional padlock pendants showing wear marks, biting down on vintage-style pink microphone with chrome mesh grill and metallic finish showing tooth pressure indentation, wearing distressed white leather motorcycle jacket with realistic hot pink paint splatter drips and cracking edges over cropped pink ribbed top, pink leather fingerless gloves with hand-formed creases and silver pyramid studs plus adjustable chrome buckles, heavy silver chain bracelets with dangling padlock charms creating weight sag, soft diffused pink neon lighting from upper left at 45 degrees casting subtle cheek shadows, shallow depth of field with blurred magenta-hued urban cityscape background showing bokeh light circles, hyper-detailed digital painting technique with visible brushwork in hair, high-fashion editorial photography aesthetic, glossy reflective chrome surfaces with environment reflections, intricate worn metal textures with micro-scratches, subsurface skin scattering on earlobes and nostrils, 8k resolution, volumetric light rays through atmosphere, --ar 9:16 --style raw --s 750

Why Physics Beats Adjectives in This AI Prompt

Are You Still Writing 'Realistic Skin' in Your Cyberpunk Singer Prompt? That is the mistake. You are giving the model a quality category, not a physical fact. Its training data links 'realistic skin' to retouched beauty images, so the statistical average comes back smoothed and plastic. The AI prompt obeys labels, not intentions. To get pores, you must describe pores. The same error happens when you write 'detailed' or 'high quality' instead of naming the material, light, and surface behavior.

Readers often treat adjectives as neutral instructions. They are not. 'Oxidized silver' is a chemical state: it changes how metal reflects light, adding darker recesses and uneven patina. 'Tooth pressure indentation' on the microphone is a physics interaction, not decoration. It anchors the prop to the body and forces contact shadows. A cyberpunk singer AI prompt becomes photorealistic when every phrase has a mechanical job. Adjectives ask the model to judge. Physics tells it what to render.

The mistake repeats with lighting. 'Moody neon' sounds useful, but it is vague. 'Magenta rim light from stage-left, cyan spill on jaw, sweat specular on upper lip' gives the model measurable events. Why does this matter? Because photorealistic portraiture depends on cause and effect: light hits surface, surface reflects, shadow falls. If your AI prompt only lists vibes, the model averages them. If it lists physical properties, the render has something concrete to solve. That is the whole shift from adjectives to physics.

Lighting as a Coordinate System, Not a Mood

The most common failure in neon-style portraits is treating neon as a color. "Neon lighting" with no further specification causes the model to tint the entire scene pink — objects become pink-colored rather than pink-lit, shadows disappear, and the image flattens into monochrome.

The breakthrough comes when you stop treating neon as a color and start treating it as a lighting condition. This prompt specifies "soft diffused pink neon lighting from upper left at 45 degrees casting subtle cheek shadows." Each clause does distinct work. "Soft diffused" sets the light quality — large source, gradual shadow edges, no harsh specular blows on skin. "Upper left at 45 degrees" establishes a coordinate, which lets the model place every downstream lighting decision consistently: the highlight on the chrome sunglass frame, the shadow under the jaw, the reflection environment in the mirrored lenses. "Casting subtle cheek shadows" explicitly requests the shadow side, preventing the flat, shadowless look that plagues AI portraits.

The problem becomes clear when you consider what happens without these clauses: the model has no reason to prefer one lighting geometry over another, so it averages across all plausible setups — and the average of all lighting directions is no direction at all.

For comparison, the ultra-realistic Joker portrait prompt uses the same directional-lighting discipline with a harder source, and the grunge street art portrait shows what happens when diffused urban light is pushed toward texture-heavy environments instead of skin.

Material Stacking: Why the Accessories Read as Real

This portrait carries an unusually high accessory load — earrings, chains, padlocks, gloves, studs, buckles, sunglasses, bandana — yet it doesn't collapse into visual noise. The reason is that each item carries a material state and a physics cue.

Consider the chain bracelets: "heavy silver chain bracelets with dangling padlock charms creating weight sag." "Weight sag" is the operative phrase. It tells the model that gravity acts on these objects, so chains drape and pull rather than hovering. Similarly, the gloves have "hand-formed creases" — creases that exist because a hand is inside them — and the jacket shows "cracking edges" on its paint splatter, meaning the paint has age and the leather flexes.

Without these interaction cues, accessories render as clip-art: visually present but physically disconnected. The model doesn't know objects must respond to gravity, contact, and wear unless you say so, because much of its training imagery — especially product renders and game art — features exactly that weightless, pristine look.

The bandana deserves separate mention: "crisp white paisley print and 'LOVE' typography wrapped tight across forehead." Text in AI images is notoriously unstable, but short strings in high-contrast placements (white on magenta, centered on a flat fabric plane) survive far better than long phrases. If typography matters to your concept, keep it under five characters and place it on a flat, front-facing surface.

Skin Optics: The Subsurface Scattering Clause

The phrase "subsurface skin scattering on earlobes and nostrils" is the most technically sophisticated element in the prompt. Subsurface scattering is the phenomenon where light enters translucent tissue, bounces internally, and exits with a warm, diffused glow. It occurs only where skin is thin enough — earlobes, nostrils, fingertips backlit by a source.

Naming the phenomenon and the anatomy matters. "Subsurface scattering" alone often gets ignored or applied uniformly, producing a waxy all-over glow. Pinning it to earlobes and nostrils gives the model anatomically correct targets, and because this portrait is lit from the upper left at close range, those thin-tissue areas are exactly where the effect would physically occur. The result is skin that reads as translucent organic matter rather than opaque painted surface — the single biggest separator between AI-looking and photograph-looking portraits.

This technique transfers directly to other portrait work. The neo-expressionist portrait guide applies the same principle in a painterly register, where scattering is simulated through warm underpainting rather than optics.

Parameter Logic: --ar 9:16, --style raw, --s 750

The parameters are tuned, not decorative. The 9:16 vertical ratio suits an extreme close-up portrait — it crops tight on the face and upper torso while leaving vertical room for the dreadlock volume above. A 16:9 or square crop would either waste frame space or force the composition wider than the subject warrants.

--style raw is critical here. Midjourney's default styling applies its own beautification pass: smoother skin, softer contrast, more conventional attractiveness. For a portrait whose entire value proposition is pores, sebum, oxidation, and wear, that default pass actively destroys the prompt's intent. Raw mode defers to your literal text.

--s 750 (stylize) sits in a deliberate middle band. At low stylize (50–150), the "hyper-detailed digital painting technique with visible brushwork in hair" clause weakens — you get straighter photorealism and lose the painterly hair texture. At very high stylize (900+), the model's artistic interpretation starts overriding physical specifications like the 45-degree light direction. 750 keeps the brushwork alive while respecting the physics. You can verify this behavior yourself on Midjourney by holding the prompt fixed and sweeping stylize values; the lighting coherence degrades noticeably above 900.

Adapting This Framework

The transferable structure here is: subject + skin physics + material states + interaction cues + directional light + targeted scattering + tuned parameters. Swap the subject and the structure holds. A cyberpunk mechanic, a futuristic dancer, a dystopian mercenary — each becomes a matter of replacing the wardrobe and prop vocabulary while keeping the physical specification density constant.

If you want to experiment, try varying one axis at a time: change the light temperature from pink neon to cold cyan and observe how the chrome reflections and skin scattering shift; or swap the 45-degree key for a low under-light to see how shadow placement restructures the face. Tools like Leonardo AI handle the same prompt with different model biases, which makes cross-platform comparison a useful way to learn which clauses are model-sensitive and which are universal. The next step in your workflow is building a personal library of these physical specification phrases — pores, oxidation, weight sag, pressure indentation — so you can compose new prompts from proven physical vocabulary rather than reinventing adjectives that don't render.