Research map / Video / Video generation
TriMotion: Modality-Agnostic Camera Control for Video Generation
Seunghyun Shin, Song Jifei, Wooseok Jeon, Hae-Gon Jeon, Jiankang Deng
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 latent motion consistency loss L_motion (ablation) | TriMotion (pose, V2V) · Camera-controlled V2V (Koala-36M + RealEstate10K) | Rotation error: 1.2307 -> 0.9488 | Improved |
| add latent motion consistency loss L_motion (ablation) | TriMotion (pose, V2V) · Camera-controlled V2V (Koala-36M + RealEstate10K) | Translation error: 4.2986 -> 3.2356 | Improved |
| add latent motion consistency loss L_motion (ablation) | TriMotion (pose, V2V) · Camera-controlled V2V (Koala-36M + RealEstate10K) | FVD: 249.85 -> 221.59 | Improved |
| condition on target embedding only instead of concatenating source and target (ablation) | TriMotion (pose, V2V) · Camera-controlled V2V (Koala-36M + RealEstate10K) | Rotation error: 2.1250 -> 0.9488 | Improved |
| condition on target embedding only instead of concatenating source and target (ablation) | TriMotion (pose, V2V) · Camera-controlled V2V (Koala-36M + RealEstate10K) | FVD: 193.56 -> 221.59 | Hurt |
| condition on target embedding only instead of subtraction (ablation) | TriMotion (pose, V2V) · Camera-controlled V2V (Koala-36M + RealEstate10K) | Translation error: 9.6904 -> 3.2356 | Improved |
| condition on pose instead of text input | TriMotion (I2V) · Camera-controlled I2V (Koala-36M + RealEstate10K) | Rotation error: 1.6875 -> 1.0113 | Improved |
| condition on video instead of text input | TriMotion (V2V) · Camera-controlled V2V (Koala-36M + RealEstate10K) | Rotation error: 1.6030 -> 1.1795 | Improved |
| TriMotion pose vs TrajectoryCrafter on rotation error | TriMotion (V2V) · Camera-controlled V2V (Koala-36M + RealEstate10K) | Rotation error: 0.6954 vs 0.9488 | Hurt |
| TriMotion pose vs CamCloneMaster on FVD (I2V) | TriMotion (I2V) · Camera-controlled I2V (Koala-36M + RealEstate10K) | FVD: 312.13 vs 317.85 | Hurt |
Closest papers by meaning
- Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance (NeurIPS 2025 · similarity 0.88)
- Training-free Camera Control for Video Generation (ICLR 2025 · similarity 0.87)
- Auteur: Language-Driven Cinematographic Framing for Human-Centric Video Generation (NeurIPS 2026 · similarity 0.86)
- DynaTokens: Teaching Dynamics to Camera-Controlled Video Models at Test Time (NeurIPS 2026 · similarity 0.86)
- DiT4DiT: Jointly Modeling Video Dynamics and Actions for Generalizable Robot Control (NeurIPS 2026 · similarity 0.86)
- GEN3C: 3D-Informed World-Consistent Video Generation with Precise Camera Control (CVPR 2025 · similarity 0.86)
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.
