C++ Path Tracer

Our first hurdle was building a Turner Whitted Ray-Tracing Algorithm capable of rendering 3D scenes featuring spheres, triangles, planes, and multiple light sources by employing the Blinn Phong shading model and other ray tracing practices such as global illumination (for reflections and refraction) as well as hard shadows.

Below you can see the different stages of development as we started to implement basic geometry intersection shadow casting and shading, finally adding secondary rays to get reflections for specular materials.

To increase realism and make the quality of the resulting images a little better we implemented anti-aliasing (notice how each pixel is visible before we apply it and how much smoother it is afterward) as well as a depth of field effect that mimics the way real cameras work. Both of these effects are achieved by introducing jittering during the ray’s sampling process. 

In the first image, we have a very hard shadow, this is because we are using a point light from which only one ray is cast. In real life, light sources have a surface like any other object (light bulbs for instance) this means that by modelling our lights as a set of multiple points we can instantly get much better shadows with, of course, a small performance penalty. 

This video documents the results we obtained, shows detailed comparisons between the same scenes rendered with different amounts of samples and presents a benchmark with and without acceleration structures.

Excited with the results we obtained previously we decided to delve into implementing the Monte Carlo Path Tracing Algorithm employing the Blinn global illumination with BRDFs allowing us to achieve diffuse and specular materials, reflections, and refraction.

Just like before we present the following video in which we document the results obtained, show detailed comparisons, and present a benchmark with and without the acceleration structure (BVH in this case).