daVinci MagiHuman is a 15B parameter omni video generation model — the new open-source king on par with WAN 2.5. Generates high-quality AI videos from text prompts with optional audio input. Supports digital humans, talking heads, and general video generation.