I wrote this program to learn about the basics of raytracing and 3d graphics. It's written in C++, and the only external libraries that it uses are SDL as backend for rendering, and ImGUI for, well, the GUI.
- Ray-sphere and axis-aligned rectangle intersections. I also implemented ray-triangle intersections using barycentric coordinates, but I'd like to implement a .obj file parser first before committing.
- Matte and specular reflections.
- Bounding Hierarchy Volume data structure to accelerate large scenes with many non-visible objects.
- Arbitrary rays based on a mix of pdf values and ray generation is supported
- For example, this program uses a mix of direct lighting and different reflection models to converge faster
Caustics and specular with 2000 samples per pixel.
The Cornell Box rendered at 2000 samples per pixel.
Implementation of fog
Example scene rendered using 1000 samples per pixel. Shows a metallic, glass, and matte material on top of a checkerboard pattern floor.
Example of the GUI of the program.
Resources used:
- https://raytracing.github.io/
- https://graphicscodex.courses.nvidia.com/
- https://lazyfoo.net/tutorials/SDL
- https://www.pbr-book.org/
Requirements for building:
- OpenMP
- SDL (Simple DirectMedia Layer)
- Eigen
This is only tested on Arch Linux, and it is not guaranteed to run on other Unix platforms. I haven't ported the program to Windows yet.