Research map / Video / Video-language understanding
TiViBench: Benchmarking Think-in-Video Reasoning for Video Generation
Harold Haodong Chen, Disen Lan, Wen-Jie Shu, Qingyang Liu, Zihan Wang, Sirui Chen, Wenkai Cheng, Kanghao Chen, Hongfei Zhang, Zixin Zhang, Rongjin Guo, Yu Cheng and 1 more
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 test-time preference prompt optimization (two samples per round, VLM critiques) | HunyuanVideo-I2V · TiViBench overall | Pass@1 (%): 4.03 -> 10.25 | Improved |
| add test-time preference prompt optimization | Wan2.1-I2V-14B · TiViBench overall | Pass@1 (%): 8.40 -> 18.15 | Improved |
| add single-pass pre-inference prompt rewriting | Wan2.1-I2V-14B · TiViBench overall | Pass@1 (%): 8.40 -> 10.76 | Improved |
| add single-pass post-inference prompt rewriting | Wan2.1-I2V-14B · TiViBench overall | Pass@1 (%): 8.40 -> 12.10 | Improved |
| add pre-inference prompt rewriting | HunyuanVideo-I2V · TiViBench overall | Pass@1 (%): 4.03 -> 4.71 | Improved |
| use prompt optimized for HunyuanVideo on Wan2.1 instead of its own optimized prompt | Wan2.1-I2V-14B · TiViBench | Pass@1 (%) | Hurt |
| sample five times (Pass@5) instead of one (Pass@1) | Wan2.1-I2V-14B · TiViBench overall | accuracy (%): 8.40 -> 15.29 | Improved |
| sample five times (Pass@5) instead of one | Wan2.2 · TiViBench overall | accuracy (%): 9.41 -> 16.47 | Improved |
| use commercial instead of open-source model (Sora 2 vs Wan2.2) | video model · TiViBench overall | Pass@1 (%): 9.41 -> 27.90 | Improved |
Closest papers by meaning
- CollabVR: Collaborative Video Reasoning with Vision-Language and Video Generation Models (NeurIPS 2026 · similarity 0.93)
- OpenCoF: Learning to Reason Through Video Generation (NeurIPS 2026 · similarity 0.92)
- VideoReasonBench: Can MLLMs Perform Vision-Centric Complex Video Reasoning? (ICLR 2026 · similarity 0.91)
- When Thinking Drifts: Evidential Grounding for Robust Video Reasoning (NeurIPS 2025 · similarity 0.91)
- Thinking with Video: Video Generation as a Promising Multimodal Reasoning Paradigm (CVPR 2026 · similarity 0.91)
- SciVideoBench: Benchmarking Scientific Video Reasoning in Large Multimodal Models (ICML 2026 · similarity 0.91)
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.
