Lizhuang Ma

Achieving state-of-the-art editing fidelity, CIME seamlessly balances motion change with structural consistency, revolutionizing text-driven human motion editing.

Even the most visually stunning video generation models struggle to maintain character continuity across shots, revealing a critical gap in current evaluation methods.

Visual in-context learning transforms video editing by seamlessly integrating visual cues with textual instructions, achieving state-of-the-art results.

Disentanglement and attribute binding are the real bottlenecks in multi-reference image generation, not scene composition, with top models still struggling to achieve high fidelity.

Fatigue can transform character animation, leading to more natural and robust movements in physics simulations.