Volumetric Cloud Rendering Techniques and Applications in Real-Time Game Environments
David FLORES
Software Development
Volumetric Cloud Rendering Techniques and Applications in Real-Time Game Environments
For my thesis, I developed a real-time volumetric cloud rendering system using compute shaders and physically based lighting models. My goal was to create a dynamic and visually realistic cloud system that could respond to changing lighting conditions in real time, all while maintaining high performance. The system included features such as octree-based voxel storage, ray marching through 3D volumes, Beer- Lambert light extinction, Henyey-Greenstein scattering, custom shadow maps, and a god ray rendering pass. I also built an interactive IMGUI debug interface to adjust parameters such as cloud density, sun position, and lighting coefficients at runtime.
I chose this project because I have a strong interest in real-time rendering techniques, volumetric effects, and advanced shader programming. I wanted to challenge myself by building a system that integrated both technical depth and visual quality, while also
solving practical performance challenges that come with volumetric rendering. The project took approximately four months to complete, from initial research and prototyping through optimization and final integration. I developed skills in GPU programming, realtime ray marching, memory optimization, physically based rendering, and debugging complex shader-based systems. I also deepened my understanding of system integration, performance profiling, and visual fidelity balancing.
This project is highly relevant to the industry. Volumetric clouds are increasingly important for creating immersive environments in open-world games, flight simulators, and large-scale outdoor scenes. A lightweight, flexible, and scalable cloud system such as mine could be adapted for dynamic weather systems, time-of-day transitions, or visual storytelling, while balancing visual richness and real-time performance.
68 SOFTWARE DEVELOPMENT