1 paper
Pengfei Zhang, Pinxin Liu, Pablo Garrido +2
Current human motion synthesis frameworks rely on global action descriptions, creating a modality gap that limits both motion understanding and generation capabilities. A single co…