Exploring the Evolution of Physics Cognition in Video Generation: A Survey

arXiv 25.03 2025 58.9 survey

TLDR

A survey on physics cognition in video generation, proposing a three-tier taxonomy from basic perception to world simulation.

Reasoning

Strengths: comprehensive taxonomy and systematic overview of physical cognition in video generation. Weaknesses: lacks empirical evaluation or new experiments; as a survey, it does not provide novel results.

Read-first score

Read-first score 58.9, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 31.

Recency 8%

86.7

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

Reproducibility 25%

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

Methodology quality 25%

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

Topical relevance 42%

44.3

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

FrontierReproducibility anchor

Rank sensitivity

Stability: volatile; rank range: 363.

Keyword Scores

video world model

world model

world simulator

world dynamics prediction

generative world model

interactive world model

model-based reinforcement learning world model

Deep Analysis

Innovations

Proposal of a three-tier taxonomy for physical cognition in video generation (basic schema perception, passive cognition of physical knowledge, active cognition for world simulation)
Adoption of a cognitive science perspective to organize the evolution of physics cognition in video generation
Comprehensive survey covering architecture designs, applications, benchmarks, and future pathways

Methodology

This survey systematically reviews video generation literature from a cognitive science perspective, proposing a three-tier taxonomy to categorize methods based on their level of physical cognition. It covers state-of-the-art methods, classical paradigms, and benchmarks, and identifies key challenges and future research directions.

Key Results

The survey proposes a three-tier taxonomy for physical cognition in video generation and identifies key challenges and potential pathways for future research, aiming to guide the development of physically consistent video generation paradigms.

Resources

Paper Code