Skin Texture Prompt for Less Oily AI Video Portraits

AI portraits can look greasy when broad quality words are paired with strong highlights. A more useful approach is to describe skin as a material: pore scale, micro-variation, gloss response, and surface condition.

The controlled test

This five-second MiniMax H3 test used the same fictional white American man, charcoal shirt, apartment, window light, framing, and face turn in both videos. The control used an ordinary natural-skin description. The technique run added fine pore texture, subtle micro-variation, low specular response, a matte-to-satin surface, and controlled highlights. Skin wording was the only variable.

What the frames show

Both clips preserved identity, lighting, and the medium close-up composition. The control looked natural but carried stronger highlights across the forehead and nose. The constrained version showed clearer pores and microtexture with slightly calmer highlights. The difference was visible but modest, so this is a limited positive observation rather than a universal fix for oily-looking skin.

A practical prompt pattern

Keep the character and lighting fixed. Then specify: fine pore texture, subtle micro-variation, low specular response, matte-to-satin surface, and controlled highlights. Add a negative boundary such as no wet look or no plastic skin only when it supports the same material goal. Avoid stacking vague quality adjectives that can amplify shine.

Where it helps

The method is most useful in portraits and close shots where skin occupies enough pixels to be judged. It is less useful in distant shots, heavily stylized characters, or scenes where the light itself is intentionally glossy.

Evidence boundary

The conclusion comes from one shared-reference A/B pair on MiniMax H3. It supports testing explicit material language, but it does not establish the same effect across other models, skin types, lighting setups, or shot scales.