This project features an interactive 2D smoke simulation that runs in real-time and integrates with a live audio analyser to produce a music visualisation. The simulator is implemented on the GPU to maximise performance. Audio input devices can be analysed to visualise live music. An FFT function is used to generate the frequency distribution of the audio streams. The resulting visual effect produced can be used by visual DJs to visualise music in a live setting such as a concert or musical performance. The effect can also be used by artists to aid in the creation of music videos for prerecorded audio.

The project was completed as part of my Honours degree in Computer Graphics at Victoria University of Wellington.

Hosted on Github here: https://github.com/EmperorJack/smokey-bbq

https://drive.google.com/open?id=0B5YqYt0mLzblQmliUDZ6Q0I5dzg

A demonstration of the smoke simulator, audio analyser and integration is available at: https://www.youtube.com/watch?v=Tsm45GY3AT4

A demonstration of the different compositions is available at: https://www.youtube.com/watch?v=Wim8OpGADB4

A comparison of linear and logarithmic frequency scaling is available at: https://www.youtube.com/watch?v=qFD48M9VfAI

• OpenGL 3.3 / GLEW
• GLFW - Window management
• GLM - Vector and matrix math
• Portaudio - Low-level audio
• KissFFT - FFT function
• Imgui - User interfaces

CMake files provided. The implementation has been tested on:

• Windows (MinGW)
• Linux
• Mac (LLVM - Clang)

Three user interface windows are provided. They can be shown or hidden with the following key bindings:

• Composition Manager GUI - "C" key
• Smoke Simulation GUI - "S" key
• Audio Analyser GUI - "A" key

Compositions define an integration between the smoke simulation and audio analyser to produce a visualisation. The current composition can be switched in the composition manager GUI. The following compositions are provided:

• Horizontal Spectrum
• Circular Spectrum
• RGB Spectrum

At the moment the video settings can only be adjusted in main.hpp by configuring the FULL_SCREEN, BORDERLESS, VSYNC, SCREEN_WIDTH and SCREEN_HEIGHT declarations.

The grid resolution can be adjusted in smoke_simulation.hpp by configuring the GRID_SIZE constant. The GPU implementation is enabled by default, if you get stuck you can switch to the CPU implementation by using the "G" key.

The sample rate, sample size and number of frequency bands can be adjusted in audio_analyser.hpp by configuring the SAMPLE_RATE, SAMPLE_SIZE and NUM_BANDS constants.

The audio analyser works by analysing a specified input device such as a microphone. This means you can plug in any audio feed into your microphone jack. The input device can be selected in the audio analyser GUI. If you want to visualise the output sound of your PC then you can use one of the following methods based on your OS to route that audio to an input device:

• Windows - Stereo mix
• Linux - PulseAudio
• Mac - Soundflower