Multi-threaded Ray Tracer

This is a multi-threaded ray tracer based on the first book of Ray Tracing in One Weekend book series by Peter Shirley, Trever D Black and Steve Hollasch. You can find all the books here.

The original book focuses on core concepts and a naive implementation, intentionally leaving room for optimization. I have extended the codebase to support CPU-based multi-threading, achieving over 10x performance improvement on a 16-core CPU.

Key Features

• Multi-threaded Rendering: Uses std::for_each with std::execution::par to saturate all available CPU cores.
• Primitives: Sphere intersection math.
• Materials:
  • Lambertian: Matte textures with accurate light diffusion.
  • Metal: Reflective surfaces with adjustable fuzziness.
  • Dielectric: Refractive materials (glass, water) with Schlick approximation.

Prerequisites

To build this project, you will need:

1. C++ Compiler: Must support C++17 (e.g., MSVC, GCC, Clang).
2. CMake: Version 3.23 or higher.
3. Intel TBB (Thread Building Blocks): Essential for <execution> support on many compilers (especially MinGW).
   • MinGW users: Install via pacman -S mingw-w64-x86_64-tbb

Build & Run

1. Clone the repository

   git clone https://github.com/ujjwalvishwakarma2006/multi-threaded-ray-tracer
   cd multi-threaded-ray-tracer

2. Configure (Ninja generator recommended for speed)

   cmake -B build -G Ninja

3. Build

   cmake --build build

4. Render The application outputs PPM image data to standard output (stdout).

   # Windows
   .\build\RayTracer.exe > image.ppm

   To see the time taken for rendering, you can use:

   .\render_image.ps1 image.ppm

Future Roadmap

• GPU Acceleration (CUDA/OptiX)
• Real-time viewport
• Complex mesh support (Triangles/OBJs)

---

This project is hand-written and maintained by me.