WorldCraft: From Camera Navigation to Object Manipulation in Interactive Video World Models

arXiv 2026 55.8 method, system

TLDR

WorldCraft extends interactive video world models from camera navigation to object-level trajectory manipulation using novel trajectory control techniques.

Reasoning

The paper addresses a clear limitation of existing video world models (only camera control) and proposes a framework with three novel components (NWT, SP-LoRA, TASP). However, the abstract lacks mention of real-world experiments or benchmarks, making it unclear if the method is empirically validated.

Read-first score

Read-first score 55.8, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 46.

Recency 6%

100

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

Citation impact 18%

83.5

Uses OpenAlex-shaped citation metadata as a bibliometric attention signal, separate from paper quality. citation_normalized_percentile=0.83548946

Topical relevance 29%

65.7

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=evaluation,experiment

Reproducibility 18%

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

Citation velocity 12%

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

Field roles

FoundationFrontierBridge

Rank sensitivity

Stability: volatile; rank range: 363.

Keyword Scores

interactive world model

video world model

world model

generative world model

world dynamics prediction

world simulator

model-based reinforcement learning world model

Deep Analysis

Innovations

Expanding interactive video world models from camera navigation to object-level trajectory actions
Normalized World Trajectory (NWT) for camera-invariant representation of user-drawn motion
Spatial-Pathway LoRA (SP-LoRA) for injecting world-space signal into the model's spatial-control pathway
Trajectory-Anchored State Persistence (TASP) for refreshing autoregressive memory after trajectory-conditioned generation

Methodology

WorldCraft uses a trajectory-centric control pipeline. First, Normalized World Trajectory (NWT) represents user-drawn motion in a camera-invariant world coordinate system and re-projects it under the current camera pose. Then, Spatial-Pathway LoRA (SP-LoRA) injects this world-space signal through the model's spatial-control pathway. Finally, Trajectory-Anchored State Persistence (TASP) refreshes autoregressive memory after trajectory-conditioned generation to maintain object state across off-camera excursions.

Key Results

WorldCraft enables accurate object control, preserves the video-based world model's camera fidelity under camera-only evaluation, and maintains object state across long autoregressive rollouts with off-camera excursions.

Resources

Paper