Beyond Pixel Histories: World Models with Persistent 3D State

arXiv 26.3 2026 52.9 method

TLDR

PERSIST introduces a world model with persistent 3D state, improving spatial memory and 3D consistency for interactive video generation.

Reasoning

The paper presents a novel paradigm that explicitly maintains a latent 3D scene, addressing key limitations of prior interactive world models. Strengths include clear methodology, quantitative and qualitative evaluations, and novel capabilities like single-image 3D synthesis. Weaknesses are not evident from the abstract alone, but the approach may have computational overhead or scalability concerns not discussed.

Read-first score

Read-first score 52.9, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 56.

Recency 6%

100

Uses a gentle age decay so recent papers surface without erasing older foundations. 2026

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,result

Reproducibility 18%

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

Citation impact 18%

34.9

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.34909398

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: 484.

Keyword Scores

world model

interactive world model

generative world model

world simulator

video world model

world dynamics prediction

model-based reinforcement learning world model

Deep Analysis

Innovations

Introduces PERSIST, a world model paradigm that simulates the evolution of a latent 3D scene (environment, camera, and renderer) for persistent spatial memory and consistent geometry.
Enables synthesizing diverse 3D environments from a single image.
Supports fine-grained, geometry-aware control over generated experiences via environment editing and specification directly in 3D space.

Methodology

PERSIST models the world by simulating the evolution of a latent 3D scene comprising environment, camera, and renderer. This allows synthesis of new frames with persistent spatial memory and consistent geometry, moving beyond pixel-level histories to a 3D state representation.

Key Results

Quantitative metrics and a qualitative user study demonstrate substantial improvements in spatial memory, 3D consistency, and long-horizon stability over existing methods, enabling coherent, evolving 3D worlds.

Resources

Paper