content-engine-dna -- Visual DNA Compiler
The Visual DNA Compiler transforms raw visual assets (photos, mood boards, reference videos, brand guidelines, character references) into persistent, structured identity files. Every downstream generation session references these compiled files instead of re-deriving visual intent from scratch.
The core insight comes from two convergent patterns:
- Karpathy's LLM Wiki -- treat visual identity like source code. Raw assets are the source. The LLM
[Description truncada. Veja o README completo no GitHub.]