Surfaces That Don’t Lie: Prompting Skin, Glass, Fabric, Metal, and Plastic

The uncanny valley isn’t about “style”—it’s about surfaces

Across this series, realism has come from getting the camera physics right: lens logic, exposure, lighting, imperfection. This part adds a different lever—the one that makes AI images look “almost” human and then quietly fail.

Surfaces give AI away because they’re where multiple physical systems overlap:

• light propagation (reflection, refraction, scattering)
• micro-geometry (roughness, pores, weave, scratches)
• material-specific spectra (skin tones, metal coloration, dielectric behavior)
• edge behavior (how highlights roll off, how transitions break)

When prompts don’t specify those behaviors, models improvise. And improvisation shows up first at material boundaries.

---

The five surface families (and the prompt handles they need)

Think of each material as a different kind of light problem. Your prompt should “name the physics,” not just the noun.

1) Skin: subsurface scattering + specular separation

Skin realism breaks when the image treats faces like plastic, satin, or overly perfect ceramic.

Use prompt handles that steer three things:

• subsurface scattering (SSS): skin should glow softly under highlights
• epidermal specular: bright highlights sit on top, not uniformly through the face
• micro-roughness + pores: highlights should not be razor-smooth

Keywords that help (use sparingly, in context):

• “subsurface scattering”
• “epidermal specular highlights”
• “soft diffuse glow under specular”
• “skin microtexture, fine pores”
• “natural skin translucency”

Common failure pattern: prompts say “realistic skin” but don’t say layered. The model then paints a flat approximation.