Research map / Video / Video generation
SparseCtrl-HOI: Sparse Temporal Control for Human-Object Interaction Video Generation
Shenbo Xie, Mingrui Cai, Xu Yang, Yifei Liu, Changxing Ding
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 data augmentation (HSV jitter and morphological mask refinement) (ablation) | SparseCtrl-HOI · SparseHOI-5K | FID: 95.14 -> 89.17 | Improved |
| add TiRoPE instead of standard 3D RoPE (ablation) | SparseCtrl-HOI · SparseHOI-5K | Ti-SSIM: 0.52 -> 0.66 | Improved |
| add TiRoPE instead of standard 3D RoPE (ablation) | SparseCtrl-HOI · SparseHOI-5K | FID: 99.28 -> 89.17 | Improved |
| add TiRoPE instead of standard 3D RoPE (ablation) | SparseCtrl-HOI · SparseHOI-5K | MS-RAFT: 0.64 -> 0.45 | Improved |
| add Motion Prior Injection (ablation) | SparseCtrl-HOI · SparseHOI-5K | MS-RAFT: 0.53 -> 0.45 | Improved |
| add Motion Prior Injection (ablation) | SparseCtrl-HOI · SparseHOI-5K | FVD: 651.18 -> 629.61 | Improved |
| add Motion Prior Injection (ablation) | SparseCtrl-HOI · SparseHOI-5K | Ti-SSIM: 0.66 -> 0.66 | No clear effect |
| broadcast reference channel-wise rather than concatenating along temporal axis (ablation) | SparseCtrl-HOI · SparseHOI-5K | FID: 111.81 -> 89.17 | Improved |
| append keyframes along temporal axis rather than channel axis (ablation) | SparseCtrl-HOI · SparseHOI-5K | FID: 202.54 -> 89.17 | Improved |
| append keyframes along temporal axis rather than channel axis (ablation) | SparseCtrl-HOI · SparseHOI-5K | convergence steps: 20k vs. 8k steps | Improved |
| increase inference keyframes K from 2 to 8 (ablation) | SparseCtrl-HOI · SparseHOI-5K | FID: 115.64 -> 81.34 | Improved |
| increase inference keyframes K from 2 to 8 (ablation) | SparseCtrl-HOI · SparseHOI-5K | FVD: 795.32 -> 515.84 | Improved |
Closest papers by meaning
- SViMo: Synchronized Diffusion for Video and Motion Generation in Hand-object Interaction Scenarios (NeurIPS 2025 · similarity 0.89)
- InterAct: Advancing Large-Scale Versatile 3D Human-Object Interaction Generation (CVPR 2025 · similarity 0.88)
- Motion Prompting: Controlling Video Generation with Motion Trajectories (CVPR 2025 · similarity 0.88)
- Auto-Regressive Diffusion for Generating 3D Human-Object Interactions (AAAI 2025 · similarity 0.88)
- HOIGen-1M: A Large-scale Dataset for Human-Object Interaction Video Generation (CVPR 2025 · similarity 0.87)
- HOI-Swap: Swapping Objects in Videos with Hand-Object Interaction Awareness (NeurIPS 2024 · similarity 0.87)
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.
