Rebalancing Reference Frame Dominance to Improve Motion in Image-to-Video Models
Addresses motion suppression in image-to-video generation for practitioners seeking better motion without retraining or quality loss.
Image-to-video models produce overly static videos due to reference-frame dominance in self-attention. DyMoS, a training-free method that rebalances attention during denoising, improves motion dynamics while preserving visual fidelity across multiple backbones.
Image-to-video models often generate videos that remain overly static, compared to text-to-video models. While prior approaches mitigate this issue by weakening or modifying the image-conditioning signal, they often require additional training or sacrifice fidelity to the reference image. In this work, we identify \emph{reference-frame dominance} as a key mechanism behind motion suppression. We observe that non-reference frames in I2V models allocate excessive self-attention to reference-frame key tokens, causing reference information to be over-propagated across time and suppressing inter-frame dynamics. Based on this finding, we propose DyMoS~(Dynamic Motion Slider), a training-free and model-agnostic method that rebalances the attention pathway from generated frames to the reference frame during initial denoising steps. DyMoS leaves both the input image and model weights unchanged and introduces a single scalar parameter for continuous control over motion strength. Experiments across multiple state-of-the-art I2V backbones demonstrate that DyMoS consistently improves motion dynamics while maintaining visual quality and fidelity to the reference image.