Phantom: Subject-Consistent Video Generation via Cross-Modal Alignment
HuMo: Human-Centric Video Generation via Collaborative Multi-Modal Conditioning
Open on GitHub