Awesome World Model Hub Papers · Datasets · Projects
← Back to papers

LatticeWorld: A Multimodal Large Language Model-Empowered Framework for Interactive Complex World Generation

arXiv 25.9 2025 51.6 method, system

TLDR

LatticeWorld uses LLM and Unreal Engine 5 to generate interactive 3D worlds from multimodal inputs with physics simulation and multi-agent interaction.

Reasoning

The paper presents a novel framework combining lightweight LLMs with a game engine for interactive world generation, supporting multimodal inputs and dynamic agents. However, the abstract lacks details on evaluation metrics and real-world validation, and the cut-off text leaves results incomplete.

Read-first score

Read-first score 51.6, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 41.

Recency 8%
86.7

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

Topical relevance 42%
58.6

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

Methodology quality 25%
50

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

Reproducibility 25%
30

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=none

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 466.

Keyword Scores

generative world model
9
interactive world model
9
world model
8
world simulator
7
world dynamics prediction
4
video world model
3
model-based reinforcement learning world model
1

Deep Analysis

Innovations

  • Integration of lightweight LLMs (LLaMA-2-7B) with industry-grade rendering engine (Unreal Engine 5) for 3D world generation
  • Multimodal input acceptance (textual descriptions and visual instructions) for world creation
  • Generation of large-scale interactive 3D worlds with dynamic agents, multi-agent interaction, high-fidelity physics simulation, and real-time rendering
  • Over 90x improvement in industrial production efficiency compared to traditional manual methods

Methodology

LatticeWorld leverages lightweight LLMs (LLaMA-2-7B) alongside the industry-grade rendering engine (e.g., Unreal Engine 5) to generate a dynamic environment. It accepts textual descriptions and visual instructions as multimodal inputs and creates large-scale 3D interactive worlds with dynamic agents, featuring competitive multi-agent interaction, high-fidelity physics simulation, and real-time rendering.

Key Results

LatticeWorld achieves superior accuracy in scene layout generation and visual fidelity, and achieves over a 90x increase in industrial production efficiency while maintaining high creative quality compared with traditional manual production methods.

Tags