subject-driven

DomainShuttle introduces a method for open domain subject-driven text-to-video generation, achieving high fidelity and flexibility across in-domain and cross-domain scenarios using domain-aware modeling and dual RoPE schemes.