WorldPlay: Towards Long-Term Geometric Consistency for Real-Time Interactive World Modeling

arXiv 25.12 2025 42.7 method, system

TLDR

WorldPlay is a real-time interactive video diffusion model achieving long-term geometric consistency via novel memory and distillation techniques.

Reasoning

Strengths include a novel memory mechanism (Reconstituted Context Memory) and distillation method (Context Forcing) enabling real-time 720p video with long-term consistency. Weaknesses are limited detail on quantitative comparisons and potential limitations not discussed in the abstract.

Read-first score

Read-first score 42.7, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 56.

Recency 6%

86.7

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

Topical relevance 29%

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 18%

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

Reproducibility 18%

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

Citation impact 18%

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. cited_by_count=0

Citation velocity 12%

Citation velocity estimates citations per publication-year to reduce old-paper bias. velocity=0.00

Field roles

Frontier

Rank sensitivity

Stability: volatile; rank range: 477.

Keyword Scores

world model

interactive world model

generative world model

video world model

world simulator

world dynamics prediction

model-based reinforcement learning world model

Deep Analysis

Innovations

Dual Action Representation for robust action control from keyboard and mouse inputs
Reconstituted Context Memory that dynamically rebuilds context from past frames and uses temporal reframing to keep geometrically important long-past frames accessible, alleviating memory attenuation
Context Forcing, a novel distillation method for memory-aware model that aligns memory context between teacher and student to preserve long-range information, enabling real-time speeds while preventing error drift

Methodology

WorldPlay is a streaming video diffusion model that uses a Dual Action Representation to enable robust action control from user inputs. It employs Reconstituted Context Memory to dynamically rebuild context from past frames with temporal reframing for long-term geometric consistency. Additionally, Context Forcing is a distillation method that aligns memory context between teacher and student to preserve long-range information and enable real-time speeds.

Key Results

WorldPlay generates long-horizon streaming 720p video at 24 FPS with superior consistency, comparing favorably with existing techniques and showing strong generalization across diverse scenes.

Resources

Paper Project