3 papers
cs.CV2026
LightCtrl: Training-free Controllable Video Relighting
Yizuo Peng, Xuelin Chen, Kai Zhang +1
Recent diffusion models have achieved remarkable success in image relighting, and this success has quickly been extended to video relighting. However, existing methods offer limite…
cs.CV2025
Democratizing High-Fidelity Co-Speech Gesture Video Generation
Xu Yang, Shaoli Huang, Shenbo Xie +3
Co-speech gesture video generation aims to synthesize realistic, audio-aligned videos of speakers, complete with synchronized facial expressions and body gestures. This task presen…
cs.GR2025
DIDiffGes: Decoupled Semi-Implicit Diffusion Models for Real-time Gesture Generation from Speech
Yongkang Cheng, Shaoli Huang, Xuelin Chen +2
Diffusion models have demonstrated remarkable synthesis quality and diversity in generating co-speech gestures. However, the computationally intensive sampling steps associated wit…