Research map / Video / Video generation
T-GVC: Trajectory-Guided Generative Video Coding at Ultra-Low Bitrates
Zhitao Wang, Hengyu Man, Wenrui Li, Xingtao Wang, Xiaopeng Fan, Debin Zhao
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 trajectory guidance instead of text guidance (ablation) | T-GVC · incl. motion bitrate | BD-rate %: 31.92 -> -3.06 | Improved |
| use trajectory guidance instead of text guidance (ablation) | T-GVC · excl. motion bitrate | BD-rate %: -4.45 -> -16.84 | Improved |
| add text guidance on top of trajectories (ablation) | T-GVC · incl. motion bitrate | BD-rate %: -3.06 -> 34.97 | Hurt |
| add text guidance on top of trajectories (ablation) | T-GVC · excl. motion bitrate | BD-rate %: -16.84 -> -13.33 | Hurt |
| use semantic-aware sparse sampling instead of dense trajectories (ablation) | T-GVC · incl. motion bitrate | BD-rate %: 188.77 -> -3.06 | Improved |
| use dense trajectories instead of sparse sampling (ablation) | T-GVC · excl. motion bitrate | BD-rate %: -16.84 vs -17.88 | No clear effect |
| use semantic-aware sampling instead of random sampling (ablation) | T-GVC · incl. motion bitrate | BD-rate %: 15.05 -> -3.06 | Improved |
| use semantic-aware sampling instead of random sampling (ablation) | T-GVC · excl. motion bitrate | BD-rate %: -7.95 -> -16.84 | Improved |
| raise K_max from 5 to 15 (ablation) | T-GVC · incl. trajectory bitrate | BD-rate %: 4.12 -> -3.06 | Improved |
| raise K_max from 15 to 20 (ablation) | T-GVC · incl. trajectory bitrate | BD-rate %: -3.06 -> 5.82 | Hurt |
| use DynamiCrafter instead of SEINE as base video model (ablation) | T-GVC · LPIPS at bitrate point 1 | BD-rate %: 0.5201 -> 0.3631 | Improved |
| use DynamiCrafter instead of Ltx-Video as base video model (ablation) | T-GVC · LPIPS at bitrate point 1 | BD-rate %: 0.4959 -> 0.3631 | Improved |
Closest papers by meaning
- GVCC: Zero-Shot Video Compression via Codebook-Driven Stochastic Rectified Flow (NeurIPS 2026 · similarity 0.88)
- Wan-Move: Motion-controllable Video Generation via Latent Trajectory Guidance (NeurIPS 2025 · similarity 0.88)
- Enhancing Motion in Text-to-Video Generation with Decomposed Encoding and Conditioning (NeurIPS 2024 · similarity 0.87)
- Motion Prompting: Controlling Video Generation with Motion Trajectories (CVPR 2025 · similarity 0.87)
- Snap Video: Scaled Spatiotemporal Transformers for Text-to-Video Synthesis (CVPR 2024 · similarity 0.87)
- Generative Pre-trained Autoregressive Diffusion Transformer (NeurIPS 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.
