Look-Before-Move: Narrative-Grounded World Visual Attention in Dynamic 3D Story Worlds

2026-06-25

AI Summary

This work addresses the challenge of active camera planning in dynamic 3D story worlds, where visual attention must balance narrative intent with physical constraints. The authors propose Look-Before-Move, a novel framework that introduces a narrative-driven visual attention mechanism into camera control: it first interprets directorial intent through a semantic observation contract, then generates compliant and feasible viewpoints via Monte Carlo viewpoint search, and finally produces coherent, obstacle-avoiding, and temporally consistent motion trajectories through semantic trajectory grounding. Evaluated on a large-scale benchmark of 50 stories built with StoryBlender, the method significantly outperforms baseline approaches in subjective perceptual quality, intent alignment, and trajectory fidelity, thereby validating the critical role of the “look-before-move” active perception paradigm in cinematic camera planning.

Abstract

As embodied AI and world models increasingly operate in dynamic 3D environments, visual perception must move beyond passively interpreting given observations toward actively deciding what to observe. We study this problem through camera planning in dynamic 3D story worlds, where the camera must not only generate smooth motion, but also decide what visual evidence should be acquired before it moves. We formulate this capability as Narrative-Grounded World Visual Attention, where the camera acts as an embodied observer that determines what to observe, how to compose the observation, and how to shift attention over time under narrative intent and physical 3D constraints. To realize this capability, we propose Look-Before-Move, a camera planning framework that separates observation specification from motion execution. It first builds a Semantic Observation Contract to convert directorial intent into executable visual constraints, then performs Monte Carlo Viewpoint Search to find narrative-compliant and geometrically feasible viewpoints, and finally applies Semantic Trajectory Grounding to connect selected viewpoints into continuous, collision-aware, and temporally coherent camera motion. We further construct a dynamic 3D Story World Benchmark based on StoryBlender, covering 50 stories, 457 scenes, and 1585 shots with animated characters, semantic scene configurations, and executable 3D environments. Experiments show that our framework improves subject perception, intent consistency, and trajectory quality over representative baselines, demonstrating the importance of organizing visual attention before generating camera motion.

Problem

• visual attention
• camera planning
• dynamic 3D environments
• narrative grounding
• embodied observation

Innovation

• Narrative-Grounded Visual Attention
• Look-Before-Move
• Semantic Observation Contract
• Monte Carlo Viewpoint Search
• Dynamic 3D Story Worlds