Celerity

A multi-threaded ray-tracing engine in C++20, written using this book for inspiration and the theory behind the ideas.

A 1920×1080p scene with 487 spheres, and 500 samples per pixel.

The above picture was rendered on my machine, which has an i5-1135G7 processor with 4 cores (with SMT, so 8 logical cores). It took 36 mins 13 secs to render the image.

Building the Project

Pre-Requisites

1. A C++20 compliant compiler
   • GCC >= 11 or MSVC >= 19.30 should do.
2. CMake

Instructions

1. Clone the repository.
   • You can clone the repo by running git clone https://github.com/wermos/Celerity.git.
2. Run CMake on the repository.
   • Run cmake -S. -Bbuild -G[generator].
     • If you are on Linux, then you can ignore the -G flag (as in, not use it at all) or just use -G"Unix Makefiles". (Since this is the default, both variants have the same behavior.)
     • If you are on Windows (with Visual Studio, presumably), you can use "Ninja" or "Visual Studio 16 2019" in place of [generator].
     • If you pass in nothing, then CMake tries to use NMake as the generator.
     • For more help, you can visit the CMake docs.
   • To pass in a specific build type, you can append -DCMAKE_BUILD_TYPE=[type] to the above command. The choices for [type] are the default ones available for any CMake project: Debug, Release, RelWithDebInfo, and MinSizeRel.
   • Run cmake --build build.
     • On Linux, you can also append -- -j<num> to the end of the above command, to run the build on <num> cores. For example,

     cmake --build build -- -j4	# for 4 cores

3. Run the executable.
   • The executable will be in build/Celerity/. The name of the executable Celerity (Linux) or Celerity.exe (Windows).
4. After running, the program saves two images called pic.jpg and pic.png to the current working directory.