This paper proposes an identity-preserving text-to-video generation framework based on a Diffusion Transformer architecture. With a frequency-aware heuristic identity-preserving control scheme, it achieves a tuning-free pipeline without case-by-case...