VideoScore2: Think before You Score in Generative Video Evaluation

arXiv 2025 37.4 method

TLDR

VideoScore2 is a multidimensional, interpretable video evaluation framework trained on human-annotated reasoning traces to assess visual quality, text alignment, and physical consistency.

Reasoning

The paper presents a strong empirical contribution with a large annotated dataset, two-stage training pipeline, and improved benchmark performance, while also offering interpretable chain-of-thought rationales. However, its focus is video evaluation rather than world modeling, so claims about world models or dynamics are only indirectly related through physical-consistency assessment.

Read-first score

Read-first score 37.4, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 2.

Recency 8%

86.7

Uses a gentle age decay so recent papers surface without erasing older foundations. 2025

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=analysis,benchmark,dataset,evaluation,experiment

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=dataset,github

Topical relevance 42%

2.9

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Field roles

FrontierMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 590.

Keyword Scores

world model

video world model

world simulator

generative world model

interactive world model

world dynamics prediction

model-based reinforcement learning world model

Resources

Paper