Carlos Jiménez de Parga PhD thesis

High-Performance Algorithms for Real-Time GPGPU Volumetric Cloud Rendering from an Enhanced Physical-Math Abstraction Approach

by Carlos Jiménez de Parga, PhD. (cjimeneztau at gmail dot com)

The aim of this thesis is performing a basic framework for real-time volumetric cloud rendering with an optimum balance between realism and performance to be applied in the entry-level graphics industry. The main features that have been achieved are the following:

Use of pseudo-spheroids
Optimized noise structure
Low number of pseudo-spheroids required
Improved bounding boxes
Improved Gaussian equations
L-System fractal cloud generation
Metaballs cloud generation
Efficient lighting system with anisotropic forward scattering
Fast light pre-calculation and occlusions thanks to the NDT algorithm
Smooth 3D mesh clouds
GPGPU cloud dynamics and morphing by using CUDA parallel programming

The NIMBUS C++ framework was implemented in Visual C++ 2017 with CUDA SDK 9.2 version using OpenGL 4.5 through Freeglut, Glew and GLSL. You can download the whole project below:

Download NIMBUS framework complete project (64-bit) (GitHub version) (v1.0 Last revision: 1/1/2025)

Download NIMBUS framework complete project (64-bit) (v1.0 Last revision 1/1/2025)

Click here for complete API documentation

Warning: Read (comment/uncomment) Defines.h header file for configuration.

The bechmarked graphics card were: