pbf-simulator

A 2D fluid simulator running entirely on the GPU with OpenGL compute shaders, built by ShaderGruppa at Hackerspace NTNU. The goal is a Position Based Fluids solver (Macklin & Müller, 2013), later extended toward multiphase and gas behaviour (Macklin et al., 2014).

Status: work in progress. The repository currently contains the compute-shader harness (window, GL context, shader loading, ping-pong SSBOs); the PBF solver itself is being built on top of it.

Requirements

• A GPU and driver supporting OpenGL 4.6 (macOS is not supported — it stops at 4.1)
• CMake ≥ 3.20 and a C++17 compiler
• Ninja (recommended)
• Git (GLFW is downloaded automatically at configure time; GLAD is vendored in external/)

Windows: MinGW-w64 (e.g. the one bundled with CLion). Linux: sudo apt install build-essential cmake ninja-build libgl1-mesa-dev xorg-dev

Build and run

git clone https://github.com/hackerspace-ntnu/pbf-simulator.git
cd pbf-simulator
cmake -S . -B build -G Ninja -DCMAKE_BUILD_TYPE=Release
cmake --build build
cd build
./pbf-simulator        # pbf-simulator.exe on Windows

The build copies shaders/ next to the executable, and the program loads shaders by relative path. Always run it from the build directory.

CLion

Open the folder as a CMake project. In the run configuration, set Working directory to $CMakeCurrentBuildDir$.

Usage

• The compute shader to run is set by shaderName in src/main.cpp.
• A shader named foo_render.comp automatically gets a matching foo_update.comp run each frame on the ping-pong buffers, if one exists.
• Hold the left mouse button and drag to rotate (for shaders that use uYaw / uPitch).

References

• M. Macklin, M. Müller. Position Based Fluids. ACM TOG 32(4), 2013. doi:10.1145/2461912.2461984
• M. Macklin et al. Unified Particle Physics for Real-Time Applications. ACM TOG 33(4), 2014. doi:10.1145/2601097.2601152