https://github.com/jonsnow1807/3d-point-cloud-viewer

A high-performance 3D point cloud visualization engine built with C++17 and OpenGL, achieving 858 FPS with 1M+ points through advanced spatial optimization.
https://github.com/jonsnow1807/3d-point-cloud-viewer

3d-visualization computer-graphics cpp frustum-culling high-performance octtree opengl point-cloud

Last synced: about 1 month ago
JSON representation

A high-performance 3D point cloud visualization engine built with C++17 and OpenGL, achieving 858 FPS with 1M+ points through advanced spatial optimization.

• Host: GitHub
• URL: https://github.com/jonsnow1807/3d-point-cloud-viewer
• Owner: JonSnow1807
• License: mit
• Created: 2025-07-09T16:04:54.000Z (12 months ago)
• Default Branch: main
• Last Pushed: 2025-07-10T05:52:03.000Z (11 months ago)
• Last Synced: 2025-07-10T14:21:36.533Z (11 months ago)
• Topics: 3d-visualization, computer-graphics, cpp, frustum-culling, high-performance, octtree, opengl, point-cloud
• Language: C++
• Homepage:
• Size: 8.78 MB
• Stars: 0
• Watchers: 0
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# 3D Point Cloud Viewer

[![C++](https://img.shields.io/badge/C%2B%2B-17-blue.svg)](https://en.wikipedia.org/wiki/C%2B%2B17)

[![OpenGL](https://img.shields.io/badge/OpenGL-3.3-green.svg)](https://www.opengl.org/)

[![CMake](https://img.shields.io/badge/CMake-3.14+-red.svg)](https://cmake.org/)

[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)

[![Performance](https://img.shields.io/badge/Performance-858%20FPS-brightgreen.svg)](benchmarks/PERFORMANCE_RESULTS.md)

A high-performance 3D point cloud visualization engine built with C++17 and OpenGL, achieving **858 FPS with 1M+ points** through advanced spatial optimization.

## Features

- **High Performance Rendering**: Renders millions of points at 60+ FPS using OpenGL

- **Spatial Optimization**: Octree-based view frustum culling and LOD (Level of Detail) rendering

- **Real-time Processing**: Point cloud filtering with <100ms execution time

- **Memory Efficient**: Custom memory pooling reduces memory footprint by 40%

- **Modular Architecture**: Clean, extensible design with CMake build system

## 🏆 Key Achievements

- **14x Performance**: 858 FPS vs 60 FPS target with 1M points

- **99% Culling Efficiency**: Only renders visible points

- **Real-time Ready**: Sub-millisecond frame times

- **Scales to 2M+ Points**: Maintains 1200+ FPS

- **Memory Optimized**: 38 bytes per point (40% reduction)

## Performance Metrics

- **Rendering**: 1M+ points at 858 FPS (14x target)

- **Filtering**: <100ms for 500K point datasets

- **Memory**: 40% reduction through custom memory pooling

- **Benchmark**: 3-4x performance improvement over PCL viewer baseline

## 🎯 Performance Showcase

### Real-World Results

| Point Count | FPS | Frame Time | Points Culled |

|------------|-----|------------|---------------|

| 10,000 | 1,546 | 0.65ms | 55% |

| 100,000 | 1,000+ | <1ms | 75% |

| 500,000 | 463 | 2.16ms | 87% |

| **1,000,000** | **858** | **1.16ms** | **96%** |

| 2,000,000 | 1,207 | 0.83ms | 99% |

*Achieved 14x the target performance (858 FPS vs 60 FPS target)*

## Architecture

```

├── Core Components

│   ├── PointCloud: Efficient point data structure

│   ├── Octree: Spatial indexing for culling and LOD

│   └── MemoryPool: Custom memory management

├── Rendering Pipeline

│   ├── Renderer: OpenGL-based rendering engine

│   ├── Camera: Interactive 3D camera system

│   └── Shaders: GPU-accelerated point rendering

└── Processing Filters

    ├── OutlierRemoval: Statistical and radius-based filtering

    └── VoxelDownsampling: Point cloud decimation

```

## Quick Start

### Prerequisites

- C++17 compatible compiler

- CMake 3.14+

- OpenGL 3.3+

- GLFW, GLEW, GLM

### Build

```bash

git clone https://github.com/JonSnow1807/3D-Point-Cloud-Viewer.git

cd 3D-Point-Cloud-Viewer

mkdir build && cd build

cmake -DCMAKE_BUILD_TYPE=Release ..

# Build with all available cores

make -j$(nproc)           # Linux

make -j$(sysctl -n hw.ncpu)  # macOS

```

### Run

```bash

./PointCloudViewer                    # Generate sample data

./PointCloudViewer cloud.xyz          # Load from file

```

## Usage

### Controls

- **WASD/QE**: Navigate camera

- **Mouse**: Look around (hold Space)

- **Scroll**: Zoom

- **ESC**: Exit

### File Format

Supports XYZ RGB format:

```

# X Y Z R G B

0.0 0.0 0.0 1.0 0.0 0.0

1.0 0.0 0.0 0.0 1.0 0.0

```

## Implementation Details

### Octree Spatial Indexing

- Hierarchical space partitioning for efficient queries

- View frustum culling eliminates non-visible points

- Dynamic LOD based on distance from viewer

### Memory Pooling

- Pre-allocated memory blocks for points

- Reduces allocation overhead

- 40% memory footprint reduction

### Rendering Pipeline

1. Frustum calculation from camera matrices

2. Octree query for visible points

3. LOD selection based on distance

4. GPU buffer updates

5. Point rendering with custom shaders

## Performance Testing

This project includes comprehensive performance benchmarks:

```bash

# Generate test data (10K to 2M points)

python3 benchmarks/scripts/generate_test_data.py

# Run interactive performance tests

./benchmarks/scripts/run_performance_test.sh

# Measure octree construction time

cd build

g++ -O3 -std=c++17 ../benchmarks/src/test_octree_timing.cpp -o octree_timing

./octree_timing

```

## Future Enhancements

- [ ] Multi-threaded octree construction

- [ ] GPU-based frustum culling

- [ ] Point cloud compression

- [ ] Support for LAS/LAZ formats

- [ ] Real-time point cloud streaming

## License

MIT License - See LICENSE file for details