Massive Particle System

This example demonstrates generating and rendering millions of particles entirely on the GPU using Compute Shaders.

Overview

A compute shader generates a noise buffer of up to ~179 million points directly in GPU memory. The buffer is converted to a point cloud mesh and rendered in Points mode with a corner-color gradient texture.

Note: This example requires approximately 2 GB of available VRAM. Reduce the noise buffer size for lower-spec hardware.

Key concepts

• Compute Shaders: Generate massive amounts of data directly on the GPU without CPU involvement. Here, a noise buffer fills GPU memory with random particle positions each frame.
• Pointcloud to Mesh: Interprets the GPU buffer as XYZ coordinates and produces renderable geometry.
• Post-processing: Edge blur applies a multi-pass Gaussian blur with Sobel edge detection. Motion blur uses a persistent feedback buffer for temporal smearing.

Data flow

1. A compute shader generates the noise particle buffer on the GPU
2. Pointcloud to Mesh builds geometry from the buffer
3. A corner-color gradient provides the texture
4. Model display renders the points
5. Edge blur and motion blur ISF Shaders are applied as post-processing

Try it

Open this example to see a massive GPU-generated particle system with dynamic blur effects.

Download this example

Learn more

• Compute Shaders - GPU compute for data generation
• ISF Shaders - Interactive Shader Format for GPU effects
• LFO - Low-frequency oscillator for animation
• Model display - 3D mesh rendering process
• Graphics pipeline - How rendering works in ossia score

---