This is a sound visualization all made on GPU. Using DirectShow.net to grab samples of mp3 audio fille in real-time. Using Cloo GPU accelerated FFT on 65536 Samples in real-time to achieve frequency. Using CUDA to move particles. Movement based on CUDA Smoke Demo mixed with frequency calculated. Call to cuda kernels are made through ManagedCuda. Using DirectX 11 on SharpDX. GPU Used is an Nvidia GTX Titan. FPS slow down based only on the amount of alpha blending not on compute FFT or Cuda Smoke simulation.
This is a brute force Nbody implementation with ManagedCuda and SharpDX.
It runs on a Geforce Titan at 30fps with 61440 stars. Recorded with NVIDIA ShadowPlay. Particles are bloomed, use of a SkyBox made with SpaceScape.
AudioSmoke is a realtime 3D visual computing application for massive particle visualization. This visual remix uses the original video art from AudioSmoke as an input source for the dynamic heightmapping method from the 2 Million Pixel Experiment to obtain a new visual style. This time the visualization is based on a tesselated surface instead of a point cloud.
AudioSmoke is a realtime 3D visual computing application to visualize millions of particles under different force field conditions. CUDA is used for the parallel computation of the force field model. The particles react on sound input (FFT analysis) and can be explored interactively in 3D space. The video shows 32 million particles under different force fields and parameter settings.
By serendipity this visual computing experiment led to a visualization of colorful structures that evolve from the parallel propagation of errors on a graphics processing unit (GPU). We wanted to transfer color values in subsequent volume slices using an OpenCL kernel. Due to concurrent memory access this transfer operation was erroneous and produced interesting visual patterns in a point-based visualization. Meanwhile we fixed this “error” with another kernel but we wanted to share the initial results as a piece of visual computing art. The visual patterns reveal the structure of block-wise kernel execution on the GPU. Further information on this experiment and code snippets (PDF).
As a mathematician, I wanted to do this experiment with CUDA. Fascinated by function studies, I made an application that plots mathematical function into 3D. For animation purposes the values for a constant K are varying over time to explore the family of curves in real-time. The CUDA kernel is compiled dynamically with all parameters and function terms.
This experiments deals with music visualization. The goal was to create as many concentric circles with radius and different color that represented a specific musical frequency, from the lowest frequency (radius = 0, color = blue) at the highest frequency (radius=k, color = yellow), each pixel is connected to an element of the vector resulting from the FFT. At the end the alpha and the color were assigned in the CUDA kernel depending on the speed and position. With this, you can watch a fountain of particles reacting to music.
This CUDA experiment maps a FULL-HD (1920×1080 @ 30 frames per second, MPEG2 compression) video source into 3D space. Each frame is processed in real-time on the GPU using CUDA. Each pixel in a frame (2.073.600 pixels per frame) is scaled by its luminance value and given the original color. The camera flight is realized with a 3D space navigator in real-time. This application is written in C# using DirectX 11, CUDA.NET and DirectShow.NET libraries. Benchmarks: GPU load is about 85% (GTX 260), GPU memory controller load 25%, CPU (i7-920) is at 20%.
Experimental application in C# using CUDA for parallel computation. The grid is quite small sized, but the GPU does it’s work: 512×512 pixels are computed by a cell formula that calculates the difference between the average neighborhood and the current cell’s value. Kernel time is about 3-4 milliseconds, stressing the GPU (GTX 260) only at 2-4% load. The sound is added from an experiment for generating digital noise.
Experimental application in C# using CUDA for parallel computation. The grid is quite small sized, but the GPU does it’s work: 512×512 pixels are calculated by various sinus functions that use their output as input while cycling through time. Kernel time is about 3-4 milliseconds, stressing the GPU (GTX 260) only at 2-4% load. The sound is added from a similar experiment that uses visually generated sinus patterns.
™ is a parallel computing platform and programming model invented by NVIDIA.
provides access to CUDA functionality through .NET applications.
3D tech news, pixel hacking, data viz and 3D programming
provides a community platform and forum integrating recent developments and news on all areas and aspects of Information Visualization.
aims an easy integration of NVidia’s CUDA in .net applications written in C#, Visual Basic or any other .net language.
is a digital arts company for innovative IT services, rapid software development with focus on visual computing technologies, sophisticated IT consulting as well as fine art content creation.
– the open standard for parallel programming of heterogeneous systems
is an advanced, low-level C# library that wraps OpenGL, OpenCL and OpenAL.
is a free open source framework that enables developers to easily build DirectX applications using .NET technologies such as C#, VB.NET, and IronPython.