Research map / 3D vision / 3D generation
Rotate Your Character: Revisiting Video Diffusion Models for High-Quality 3D Character Generation
Jin Wang, Jianxiang Lu, Comi Chen, Guangzheng Xu, Haoyu Yang, Peng Chen, Na Zhang, Yifan Xu, Longhuang Wu, Shuai Shao, Qinglin Lu, Ping Luo
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 Stage III training (ablation) | RCM (Wan 2.1) · Training-stage ablation | FID: 155.91 -> 54.83 | Improved |
| add Stage III training (ablation) | RCM (Wan 2.1) · Training-stage ablation | CD: 0.2175 -> 0.0960 | Improved |
| add Stage I training (ablation) | RCM (Wan 2.1) · Training-stage ablation | FID: 56.06 -> 54.83 | Improved |
| add Stage II training (ablation) | RCM (Wan 2.1) · Training-stage ablation | FID: 58.49 -> 54.83 | Improved |
| add Stage I training (ablation) | RCM (Wan 2.1) · Training-stage ablation | CD: 0.1064 -> 0.0960 | Improved |
| add Stage II training (ablation) | RCM (Wan 2.1) · Training-stage ablation | CD: 0.1091 -> 0.0960 | Improved |
| use 3-view instead of 1-view conditioning | RCM · RCM-Hard | PSNR: 19.41 -> 23.58 | Improved |
| use 3-view instead of 1-view conditioning | RCM · RCM-Hard | SSIM: 0.89 -> 0.92 | Improved |
| train all objectives jointly without staging (ablation) | RCM (Wan 2.1) · Wan 2.1 base | viewpoint controllability | Hurt |
| inject camera pose via cross-attention or IP-Adapter instead of Camera Encoder (ablation) | RCM (Wan 2.1) · RCM | visual quality | Hurt |
| rightwards zero-padding of reference latents (ablation) | RCM (Wan 2.1) · RCM | pose canonicalization | Hurt |
| omit background augmentation (ablation) | RCM (Wan 2.1) · RCM | background entanglement | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- Epidiff — EpiDiff: Enhancing Multi-View Synthesis via Localized Epipolar-Constrained Diffusion (CVPR 2024) · better on 3 of 3 shared results
- SyncDreamer — SyncDreamer: Generating Multiview-consistent Images from a Single-view Image (ICLR 2024) · better on 3 of 3 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
- ViVid-1-to-3: Novel View Synthesis with Video Diffusion Models (CVPR 2024 · similarity 0.89)
- Animate Anyone: Consistent and Controllable Image-to-Video Synthesis for Character Animation (CVPR 2024 · similarity 0.88)
- MIMO: Controllable Character Video Synthesis with Spatial Decomposed Modeling (CVPR 2025 · similarity 0.87)
- Morphable Diffusion: 3D-Consistent Diffusion for Single-image Avatar Creation (CVPR 2024 · similarity 0.87)
- SynCamMaster: Synchronizing Multi-Camera Video Generation from Diverse Viewpoints (ICLR 2025 · similarity 0.87)
- Efficient-3Dim: Learning a Generalizable Single-image Novel-view Synthesizer in One Day (ICLR 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.
