Research map / Video / Video generation
Cosh-DiT: Co-Speech Gesture Video Synthesis via Hybrid Audio-Visual Diffusion Transformers
Yasheng Sun, Zhiliang Xu, Hang Zhou, Jiazhi Guan, Quanwei Yang, Kaisiyuan Wang, Borong Liang, Yingying Li, Haocheng Feng, Jingdong Wang, Ziwei Liu, Koike Hideki
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add AM-DiT iterative blocks (ablation) | Cosh-DiT · Co-speech gesture video (ablation) | FID: FID 84.22 -> 77.51 | Improved |
| add AM-DiT iterative blocks (ablation) | Cosh-DiT · Co-speech gesture video (ablation) | IDSim: IDSim 0.5742 -> 0.5918 | Improved |
| add spatial alignment of 2D skeleton and 3D hand mesh (ablation) | Cosh-DiT · Co-speech gesture video (ablation) | FID: FID 89.24 -> 77.51 | Improved |
| add spatial alignment of 2D skeleton and 3D hand mesh (ablation) | Cosh-DiT · Co-speech gesture video (ablation) | Conf-body: Conf-body 0.8723 -> 0.9343 | Improved |
| add spatial alignment of 2D skeleton and 3D hand mesh (ablation) | Cosh-DiT · Co-speech gesture video (ablation) | IDSim: IDSim 0.5423 -> 0.5918 | Improved |
| add hybrid motion with 3D hand guidance (ablation) | Cosh-DiT · Co-speech gesture video (ablation) | Conf-hand: Conf-hand 0.8623 -> 0.8937 | Improved |
| add hybrid motion with 3D hand guidance (ablation) | Cosh-DiT · Co-speech gesture video (ablation) | SSIM: SSIM 0.6584 -> 0.6729 | Improved |
| add AM-DiT iterative blocks (ablation) | Cosh-DiT · Co-speech gesture video (ablation) | SSIM: SSIM 0.6323 -> 0.6729 | Improved |
| use Cosh-DiT-V instead of ControlNeXt for pose-driven animation | Cosh-DiT-V · Motion animation with pose input | FID: FID 55.64 -> 24.62 | Improved |
| compare against S2G-MDD GAN-based rendering | Cosh-DiT · Co-speech gesture video (S2G-MDD evaluation data) | FVD: FVD 1201.72 -> 881.83 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- AnimateAnyone — Animate Anyone: Consistent and Controllable Image-to-Video Synthesis for Character Animation (CVPR 2024) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Co-Speech Gesture Video Generation via Motion-Decoupled Diffusion Model (CVPR 2024 · similarity 0.93)
- DiffSHEG: A Diffusion-Based Approach for Real-Time Speech-driven Holistic 3D Expression and Gesture Generation (CVPR 2024 · similarity 0.91)
- ConvoFusion: Multi-Modal Conversational Diffusion for Co-Speech Gesture Synthesis (CVPR 2024 · similarity 0.90)
- Co-Speech Gesture Video Generation with Implicit Motion-Audio Entanglement (CVPR 2025 · similarity 0.90)
- Co$^{\mathbf{3}}$Gesture: Towards Coherent Concurrent Co-speech 3D Gesture Generation with Interactive Diffusion (ICLR 2025 · similarity 0.89)
- AudCast: Audio-Driven Human Video Generation by Cascaded Diffusion Transformers (CVPR 2025 · similarity 0.89)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
