TriangleVision 📐✨

TriangleVision is a high-performance video processing tool that converts standard video (e.g., MP4) into a stylized, real-time mesh of triangles. It blends geometry and aesthetics to create a unique visual style while maintaining image detail and screen ratio.

Features 🚀

• Real-time Processing: Process webcam or video files at 30+ FPS (on modern hardware).
• GPU-Accelerated Rendering: Utilizes ModernGL for high-performance triangle rendering.
• Intelligent Point Sampling:
  • Delaunay Triangulation: Creates a smooth, connected mesh.
  • Edge Density Adaptation: Automatically adds more detail to high-complexity areas.
  • Human/Face Detection: Enhanced detail in areas containing people.
  • Motion Tracking: Increased triangle density in regions with significant movement.
• Custom .triv Codec: A custom binary format for storing triangle-based video data with high efficiency.
• Multiple Modes:
  • Realtime: Live camera or video file playback with real-time triangulation.
  • Encode: Convert standard videos into optimized .triv files.
  • Player: High-speed playback of .triv files.
  • Export: Export triangulated videos back to standard formats (MP4, MKV).
• Visual Styles: Toggle between standard triangulation, Rotoscope (ink edges + cell shading), and Heatmap modes.

How it's Implemented 🛠️

TriangleVision is built with Python and leverages several powerful libraries for performance and geometric calculations:

• OpenCV: Core image processing, feature detection (Shi-Tomasi), and human detection (HOG + Haar Cascades).
• SciPy (Delaunay): High-speed triangulation of sampled points.
• ModernGL: GPU-accelerated triangle rendering with custom GLSL shaders.
• NumPy & Numba: Vectorized operations and JIT compilation for color sampling and data processing.
• Threaded Capture: Multi-threaded video grabbing to ensure smooth input regardless of processing load.

Getting Started 🏁

Prerequisites

• Python 3.8+
• A GPU supporting OpenGL 3.3+ (for hardware acceleration)

Installation

1. Clone the repository:

   git clone https://github.com/yourusername/triangleVision.git
   cd triangleVision

2. Install dependencies:

   pip install -r requirements.txt

Usage

1. Real-time Webcam Mode

Run the tool on your webcam with default settings:

python3 main.py realtime --quality medium --human

2. Encode a Video File

Convert an MP4 to a .triv file at high quality:

python3 main.py encode input.mp4 output.triv --quality high --human

3. Play a .triv File

Enjoy the triangulated version of your video:

python3 main.py play output.triv

4. Export to MP4

Export a stylized version for sharing:

python3 main.py export output.triv finished_video.mp4 --rotoscope

CLI Arguments ⌨️

• --source: Camera ID or video file path (for realtime).
• --triangles: Manually set the target number of triangles.
• --quality: Adaptive quality presets (low, medium, high).
• --rotoscope: Enable ink edges and cell-shading aesthetics.
• --human: Enable smart detail enhancement on humans.
• --output: Save a realtime session directly to a .triv file.

License 📄

This project is licensed under the MIT License - see the LICENSE file for details.