Research map / 3D vision / 3D generation
3D-Aware Implicit Motion Control for View-Adaptive Human Video Generation
Zhixue Fang, Xu He, Songlin Tang, Haoxian Zhang, Qingfeng Li, Xiaoqiang Liu, Pengfei Wan, Kun Gai
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 |
|---|---|---|---|
| use learned implicit motion tokens instead of SMPL control (ablation) | 3DiMo · Ablation (Table 2) | FVD: 348.2 -> 297.4 | Improved |
| use cross-attention instead of channel concatenation (ablation) | 3DiMo · Ablation (Table 2) | FVD: 395.6 -> 297.4 | Improved |
| add auxiliary geometric supervision early in training (ablation) | 3DiMo · Ablation (Table 2) | FID: 51.3 -> 36.9 | Improved |
| add hand motion encoder (ablation) | 3DiMo · Ablation (Table 2) | FID: 38.1 -> 36.9 | Improved |
| add stages 2 and 3 (view-rich training) (ablation) | 3DiMo · Ablation (Table 2) | FID: 40.5 -> 36.9 | Mixed |
| add stage 3 (view-rich only) (ablation) | 3DiMo · Ablation (Table 2) | LPIPS: 0.221 -> 0.221 | No clear effect |
| add stage 3 to stage 1 and 2 (ablation) | 3DiMo · Ablation (Table 2) | FVD: 314.5 -> 297.4 | Improved |
| implicit 3D-aware motion vs SMPL-based Uni3C | 3DiMo · TikTok + internet videos (150) | FVD: 321.9 -> 297.4 | Improved |
| implicit 3D-aware motion vs MTVCrafter on pixel metrics | 3DiMo · TikTok + internet videos (150) | PSNR: 18.03 -> 17.96 | Hurt |
Closest papers by meaning
- LooseControlVideo: Directorial Video Control using Spatial Blocking (ECCV 2026 · similarity 0.90)
- AC3D: Analyzing and Improving 3D Camera Control in Video Diffusion Transformers (CVPR 2025 · similarity 0.89)
- GEN3C: 3D-Informed World-Consistent Video Generation with Precise Camera Control (CVPR 2025 · similarity 0.89)
- 3DTrajMaster: Mastering 3D Trajectory for Multi-Entity Motion in Video Generation (ICLR 2025 · similarity 0.89)
- Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance (NeurIPS 2025 · similarity 0.88)
- VideoMDM: Towards 3D Human Motion Generation From 2D Supervision (NeurIPS 2026 · similarity 0.88)
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.
