SKILL: Skin Tone Inconsistency

Version: 1.0 | Hermes Agent Failure Pattern Library

DESCRIPTION

Mastery of preventing and correcting skin tone drift across AI-generated character series, video frames, and multi-subject scenes. Skin tone inconsistency is a critical failure in commercial and ethical AI imagery — it undermines character consistency, misrepresents identity, and signals low production quality. This skill provides detection methods, reference anchoring vocabulary, and normalization workflows [^80^][^83^].

TRIGGER KEYWORDS

skin tone, skin color, skin inconsistency, tone drift, skin color change, complexion, skin tone lock, melanin, dark skin, light skin, skin consistency, tone mismatch, racial representation, complexion accuracy

CORE RULES

• Anchor skin tone with specific descriptors, not generic: "deep brown skin with warm undertones" not "dark skin"
• Lighting changes apparent skin tone: warm light adds amber, cool light adds gray — compensate in prompt
• Use the Fitzpatrick scale reference for consistency: Type I–VI with descriptive adjectives
• Place skin tone description early in prompt — especially important for Flux
• For video: "skin tone consistent across all frames, no color drift" is a required guardrail
• Multi-character scenes: name each character's skin tone separately to prevent averaging
• Environmental color cast (neon, warm sunset) must be separated from true skin tone in description