Qingxuan Wu

Text2Interact: High-Fidelity and Diverse Text-to-Two-Person Interaction Generation

DICE: End-to-end Deformation Capture of Hand-Face Interactions from a Single Image