Research map / Video / Video generation
Controllable Video Generation through Global and Local Motion Dynamics
Aram Davtyan, Paolo Favaro
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 vector quantization of actions instead of Gumbel-Softmax MLP (ablation) | GLASS LMA · W-Sprites | FVD: 304 -> 188 | Improved |
| add vector quantization instead of Gumbel-Softmax MLP (ablation) | GLASS LMA · W-Sprites | NMI_S: 0.02 -> 0.41 | Improved |
| add modified mapping T instead of plain feature difference (ablation) | GLASS LMA · W-Sprites | NMI_S: 0.17 -> 0.41 | Improved |
| add modified mapping T instead of plain feature difference (ablation) | GLASS LMA · W-Sprites | mIoU_RE: 0.83 -> 0.88 | Improved |
| add modulated convolutions for action embeddings (ablation) | GLASS LMA · W-Sprites | FVD: 173 -> 188 | Hurt |
| add modulated convolutions for action embeddings (ablation) | GLASS LMA · W-Sprites | NMI_S: 0.38 -> 0.41 | Mixed |
| feed global and local embeddings to separate RNN blocks instead of jointly (ablation) | GLASS LMA · W-Sprites | FVD: 156 -> 188 | Hurt |
| feed global and local embeddings to separate RNN blocks instead of jointly (ablation) | GLASS LMA · W-Sprites | FID: 26.3 -> 27.8 | Hurt |
| add L_RECS loss (ablation) | GLASS LMA · W-Sprites | FID: 29.3 -> 27.8 | Improved |
| add L_RECS loss (ablation) | GLASS LMA · W-Sprites | NMI_S: 0.30 -> 0.41 | Improved |
| add L_RECB loss (ablation) | GLASS GMA · W-Sprites | mIoU: 0.01 -> 0.88 | Improved |
| add L_VGG loss (ablation) | GLASS GMA · W-Sprites | mIoU: 0.89 -> 0.88 | Hurt |
Closest papers by meaning
- Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance (NeurIPS 2025 · similarity 0.88)
- CAGE: Unsupervised Visual Composition and Animation for Controllable Video Generation (AAAI 2025 · similarity 0.87)
- Motion Prompting: Controlling Video Generation with Motion Trajectories (CVPR 2025 · similarity 0.86)
- DynaTokens: Teaching Dynamics to Camera-Controlled Video Models at Test Time (NeurIPS 2026 · similarity 0.86)
- Move-in-2D: 2D-Conditioned Human Motion Generation (CVPR 2025 · similarity 0.86)
- MotionPro: A Precise Motion Controller for Image-to-Video Generation (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.
