A visual perception-guided framework for camera path planning in large-scale digital twin and AI-generated 3D scenes.

Abstract

This study investigates visual, cognitive, and spatial saliency indicators of architectural landmarks in large-scale ancient city ruins and digital twin virtual environments. It further examines users' cognitive demands during pathfinding in large, complex environments and explores how different exploration strategies interact with interact with architectural spatial layouts. We propose a virtual camera exploration method guided by visual perception principles. Comparative experiments with several representative path optimization algorithms show that the proposed method outperforms existing methods and provides effective optimization strategies. This method identifies visually engaging and aesthetically rich regions in large-scale, complex virtual environments. As a result, it improves users' exploration efficiency in such environments with multiple layers and complex structures, while enhancing their scene understanding and aesthetic experience. The proposed method has substantial practical value across multiple fields, including landscape animation, 3D heritage simulation, and virtual cultural tourism.