Xize Cheng

OmniBind integrates various data types for improved content understanding and generation.

New model improves lip synchronization and visual quality in talking head videos.

SegTalker enhances talking face videos with realistic textures and easy editing.

Wander enhances efficiency in multimodal models for better data processing.