We introduce sequence models that are equivariant to time-parameterized symmetries such as motion.

We propose RoPECraft, a training-free video motion transfer method for diffusion transformers that operates solely by modifying their rotary positional embeddings (RoPE).