Research map / Video / Video generation
ImagerySearch: Adaptive Test-Time Search for Video Generation Beyond Semantic Dependency Constraints
Meiqi Wu, Jiashu Zhu, Xiaokun Feng, Chubin Chen, Chen Zhu, Bingze Song, Fangyuan Mao, Jiahong Wu, Xiangxiang Chu, Kaiqi Huang
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 ImagerySearch on top of Wan2.1 | ImagerySearch · LDT-Bench | ImageryQA (All): 48.28 -> 57.11 | Improved |
| use ImagerySearch instead of Video-T1 | ImagerySearch · LDT-Bench | ImageryQA (All): 54.75 -> 57.11 | Improved |
| use ImagerySearch instead of EvoSearch | ImagerySearch · VBench (LDT-Bench prompts) | average: 82.08 -> 83.48 | Improved |
| add Adaptive Imagery Reward (AIR) (ablation) | ImagerySearch · VBench ablation | average: 82.11 -> 83.10 | Improved |
| add semantic-distance-aware dynamic search space (SaDSS) (ablation) | ImagerySearch · VBench ablation | average: 81.30 -> 83.10 | Improved |
| use dynamic instead of fixed search space (ablation) | ImagerySearch · VBench ablation | average: 81.22 -> 83.48 | Improved |
| increase static search-space weight from 0.5 to 0.9 (ablation) | ImagerySearch · VBench ablation | average: 81.18 -> 81.22 | No clear effect |
| use ImagerySearch instead of Best-of-N search (ablation) | ImagerySearch · VBench ablation | average: 82.64 -> 83.10 | Improved |
| use ImagerySearch instead of particle sampling (ablation) | ImagerySearch · VBench ablation | average: 81.39 -> 83.10 | Improved |
| increase inference compute (NFEs) | ImagerySearch · Test-time scaling | VideoAlign score | Improved |
| change reward weights from 0.2 to 1.2 (ablation) | ImagerySearch · VBench | TA vs MQ/VQ | Mixed |
| apply test-time search (Video-T1) on Wan2.1 | ImagerySearch · VBench (LDT-Bench prompts) | Dynamic Degree: 82.85 -> 54.05 | Hurt |
Closest papers by meaning
- TTS-VAR: A Test-Time Scaling Framework for Visual Auto-Regressive Generation (NeurIPS 2025 · similarity 0.88)
- Video-Bench: Human-Aligned Video Generation Benchmark (CVPR 2025 · similarity 0.87)
- GenIR: Generative Visual Feedback for Mental Image Retrieval (NeurIPS 2025 · similarity 0.87)
- The Devil is in the Prompts: Retrieval-Augmented Prompt Optimization for Text-to-Video Generation (CVPR 2025 · similarity 0.87)
- IDEA-Bench: How Far are Generative Models from Professional Designing? (CVPR 2025 · similarity 0.87)
- Stable Cinemetrics : Structured Taxonomy and Evaluation for Professional Video Generation (NeurIPS 2025 · 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.
