RoPECraft is a training-free method that modifies rotary positional embeddings in diffusion transformers to transfer motion from reference videos, enhancing text-guided video generation and reducing artifacts.
We propose RoPECraft, a training-free video motion transfer method for
diffusion transformers that operates solely by modifying their rotary
positional embeddings (RoPE). We first extract dense optical flow from a
reference video, and utilize the resulting motion offsets to warp the
complex-exponential tensors of RoPE, effectively encoding motion into the
generation process. These embeddings are then further optimized during
denoising time steps via trajectory alignment between the predicted and target
velocities using a flow-matching objective. To keep the output faithful to the
text prompt and prevent duplicate generations, we incorporate a regularization
term based on the phase components of the reference video’s Fourier transform,
projecting the phase angles onto a smooth manifold to suppress high-frequency
artifacts. Experiments on benchmarks reveal that RoPECraft outperforms all
recently published methods, both qualitatively and quantitatively.