Stefan Claasse

To make the ray tracer interesting, I added a few material types to show its effect. Those include: opaque, reflection and refraction. Besides those, material color was also added to change the effect.

Opaque: the opaque object within the scene is very basic, and simply traces the rays from the screen towards the light source, while also taking the material color into account to display the correct color. A Specular was also added for additional effect where the lights focus is the strongest.

Reflection: the reflective objects within the scene are a bit more interesting, as the rays don’t go directly towards the light source, but rather follow a certain angle depending on where the camera is. From there, the ray would hit a mat object and trace back to the light source. Resulting in a reflective material. Depending on how reflective a material is, it might retain some of its original color.

Refraction: the refractive objects in the scene act similar to the reflections but split the rays in 2. Where one ray would reflect and the other would refract into the refractive material. The ray that’s inside the refractive material would continue to split in two, where one ray would stay inside and the other would continue outside to trace back to the light source.

As continuous refraction is impossible to calculate, a hard cap has been added to stop the rays from refracting. Depending on the refraction value of the object, the result will also differ. As water has a different fraction value compared to glass for example.

When firing a ray from the screen towards an object in the scene back to a light source. It might happen that another object blocks this path. In these cases the pixel would turn black as it became a occluded, and is now a shadow. Although this will result in hard shadows. They could be made softer to result in more realistic shadows.

One optimization technique that I applied to my CPU ray tracer was the implementation of a Bounding Volume Hierarchy (BVH).

Using the BVH, I would turn the objects within the scene into leaves that follow back to a root. This would optimize the ray tracer by only having to check if a ray hit a single root, rather than an individual object. Thus, reducing the cycles needed to check if the rays hit something.

Although not implemented in the project, one optimization technique that I could implemented would be multithreading for different sections of the screen. Some other techniques would also include making use of SIMD to run calculations in parallel without multiple threads. And to add noise to reduce the amount of rays to be fired.