Future Dynamic 3D Reconstruction: A 3D World Model with Disentangled Ego-Motion

arXiv 2026 57.5 method

TLDR

FR3D predicts future dynamic 3D scenes by disentangling ego-motion from world dynamics, using teacher-student distillation for zero-shot generalization.

Reasoning

Strengths include explicit disentanglement of ego-motion and scene dynamics for geometric consistency, and leveraging foundation models for robust zero-shot generalization. Weaknesses are the limited prediction horizon (2 seconds) and lack of discussion on real-time performance or computational cost.

Read-first score

Read-first score 57.5, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 35.

Recency 6%

100

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

Citation impact 18%

92.9

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.92860865

Methodology quality 18%

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

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

Reproducibility 18%

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

Citation velocity 12%

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

Field roles

FoundationFrontierBridgeMethodology anchor

Rank sensitivity

Stability: volatile; rank range: 380.

Keyword Scores

world model

world dynamics prediction

generative world model

world simulator

video world model

interactive world model

model-based reinforcement learning world model

Deep Analysis

Innovations

FR3D predicts a persistent 3D latent representation for future dynamic 3D reconstruction, unlike prior works that treat the world as a sequence of image-based features.
Explicit decoupling of the 3D evolution of the scene from the agent's trajectory, treating inferred ego-motion as a latent proxy for action to resolve ambiguities between self-motion and world-motion.
Teacher-student distillation strategy that leverages the spatial 'common sense' of off-the-shelf foundation models for robust zero-shot generalization.

Methodology

FR3D is a world model that predicts a persistent 3D latent representation for future dynamic 3D reconstruction. It explicitly decouples the 3D evolution of the scene from the agent's trajectory, treating inferred ego-motion as a latent proxy for action. A teacher-student distillation strategy leverages off-the-shelf foundation models for robust zero-shot generalization.

Key Results

Extensive experiments demonstrate FR3D's strong performance for future dynamic 3D reconstruction from monocular observations across multiple datasets, even 2 seconds into the future.

Resources

Paper