Paper 2510.14847
ImagerySearch: Adaptive Test-Time Search for Video Generation Beyond Semantic Dependency Constraints
- Published
- Oct 2025
- Research lab
- Independent
- Citations
- 12
- GitHub
- 56 stars
01 In brief
Summary
The paper introduces ImagerySearch, a test-time search strategy for text-to-video generation that adapts to prompts with long-distance semantic relationships, which are rare in training data and cause performance degradation in imaginative scenarios.
ImagerySearch dynamically adjusts the inference search space (SaDSS) and reward function (AIR) based on the semantic distance between prompt entities, improving coherence and visual plausibility.
The authors also present LDT-Bench, the first benchmark for long-distance semantic prompts, containing 2,839 prompts (1,938 objects, 901 actions) and an automated evaluation protocol (ImageryQA) covering element coverage, semantic alignment, and anomaly detection.
Experiments show ImagerySearch outperforms general models (e.g., Wan2.1, Hunyuan) and existing test-time scaling methods (e.g., Video-T1, EvoSearch) on LDT-Bench, achieving an ImageryQA score of 57.11%, and on VBench, achieving an average score of 83.48%.
Ablations confirm the effectiveness of both modules and the dynamic search space.
The code and benchmark will be released.
02 From the paper
Abstract
Video generation models have achieved remarkable progress, particularly excelling in realistic scenarios; however, their performance degrades notably in imaginative scenarios. These prompts often involve rarely co-occurring concepts with long-distance semantic relationships, falling outside training distributions. Existing methods typically apply test-time scaling for improving video quality, but their fixed search spaces and static reward designs limit adaptability to imaginative scenarios. To fill this gap, we propose ImagerySearch, a prompt-guided adaptive test-time search strategy that dynamically adjusts both the inference search space and reward function according to semantic relationships in the prompt. This enables more coherent and visually plausible videos in challenging imaginative settings. To evaluate progress in this direction, we introduce LDT-Bench, the first dedicated benchmark for long-distance semantic prompts, consisting of 2,839 diverse concept pairs and an automated protocol for assessing creative generation capabilities. Extensive experiments show that ImagerySearch consistently outperforms strong video generation baselines and existing test-time scaling approaches on LDT-Bench, and achieves competitive improvements on VBench, demonstrating its effectiveness across diverse prompt types. We will release LDT-Bench and code to facilitate future research on imaginative video generation.