Neural rendering based digital twin scene high-fidelity reconstruction system

Abstract

The application discloses a high-fidelity reconstruction system for a digital twin scene based on neural rendering, comprising a high-fidelity rendering module, a structure cognition module, a principle learning module, a maintenance training module, an examination and evaluation module, a multi-person cooperation module, an editing and publishing module, a hardware adaptation module, a resource management module and a user interface module. The application introduces neural rendering technology to realize high-fidelity texture and light reconstruction of a mechanical structure, enhance the reality and immersion of a scene, support various simulation modes such as structure cognition, principle learning, maintenance training and examination and evaluation through a modular editor and a case system, meet the training needs in different scenes, support PC, VR and multi-person cooperation operation, have flexible device adaptation and interaction capability, and provide a whole-process editing tool from model import, animation editing to step arrangement, greatly reduce the production threshold of simulation content, and improve development efficiency.

Images