Research map / Video / Video generation
TGT: Text-Grounded Trajectories for Locally Controlled Video Generation
Guofeng Zhang, Angtian Wang, Jacob Zhiyuan Fang, Liming Jiang, Haotian Yang, Bo Liu, Yiding Yang, Guang Chen, Longyin Wen, Alan Yuille, Chongyang Ma
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 sparse-track tuning after dense-track training (ablation) | LACA · DAVIS | EPE: 58.01 -> 45.28 | Improved |
| add sparse-track tuning after dense-track training (ablation) | LACA · DAVIS | CLIP-T (local): 0.2394 -> 0.2447 | Improved |
| add sparse-track tuning after dense-track training (ablation) | LACA · DAVIS | CLIP-T (global): 0.3307 -> 0.3312 | No clear effect |
| add Gaussian masking to sparse-track LACA (ablation) | LACA · DAVIS | EPE: 45.28 -> 25.11 | Improved |
| add Gaussian masking to sparse-track LACA (ablation) | LACA · DAVIS | CLIP-T (local): 0.2447 -> 0.2527 | Improved |
| use dual-CFG instead of global-only guidance (ablation) | TGT · DAVIS | EPE: 91.38 -> 25.11 | Improved |
| use dual-CFG instead of local-only guidance (ablation) | TGT · DAVIS | CLIP-T (global): 0.3117 -> 0.3314 | Improved |
| use dual-CFG instead of local-only guidance (ablation) | TGT · DAVIS | EPE: 43.29 -> 25.11 | Improved |
| use dual-CFG instead of combined global-with-local guidance (ablation) | TGT · DAVIS | EPE: 53.30 -> 25.11 | Improved |
| use dual-CFG instead of combined global-with-local guidance (ablation) | TGT · DAVIS | CLIP-T (local): 0.2491 -> 0.2527 | Improved |
| add extended global+local prompts describing motion and location to Wan | Wan2.2 14B · DAVIS | EPE: 265.03 -> 180.36 | Improved |
| use TGT instead of Tora trajectory control | Wan2.1 14B · DAVIS | EPE: 47.41 -> 25.11 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Tora — Tora: Trajectory-oriented Diffusion Transformer for Video Generation (CVPR 2025) · 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
- Enhancing Motion in Text-to-Video Generation with Decomposed Encoding and Conditioning (NeurIPS 2024 · similarity 0.89)
- Trajectory attention for fine-grained video motion control (ICLR 2025 · similarity 0.88)
- Motion Prompting: Controlling Video Generation with Motion Trajectories (CVPR 2025 · similarity 0.88)
- TIV-Diffusion: Towards Object-Centric Movement for Text-driven Image to Video Generation (AAAI 2025 · similarity 0.88)
- Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance (NeurIPS 2025 · similarity 0.88)
- Unveiling the Value of Motion for Cinematic Camera Trajectories (NeurIPS 2026 · 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.
