Machine Learning-Driven Volumetric Cloud Rendering: Procedural Shader Optimization and Dynamic Lighting in Unreal Engine for Realistic Atmospheric Simulation

• DOI: 10.48550/arxiv.2502.08107
• Publication Date: 2025-01-01
• Authors: Singh, Shruti; Kumar, Shantanu
• Keywords: FOS: Computer and information sciences; Computer Science - Graphics; Graphics (cs.GR)

This study advances real-time volumetric cloud rendering in Computer Graphics (CG) by developing a specialized shader in Unreal Engine (UE), focusing on realistic cloud modeling and lighting. By leveraging ray-casting-based lighting algorithms, this work demonstrates the practical application of a dual-layered procedural noise model, eliminating the need for conventional two-dimensional (2D) weather textures. The shader allows for procedurally configured skies with a defined parameter set, offering flexibility for both artistic expression and realistic simulation. Empirical results reveal that the shader achieves an average rendering time of 35ms per frame while maintaining high visual accuracy and scene realism. Visual fidelity assessments indicate a 15% improvement in cloud realism over traditional 2D techniques, particularly in dynamic lighting scenarios. This research contributes to CG by bridging technical and aesthetic elements, enhancing real-time visual storytelling and immersion within gigital media environments.