Homework 2 - Basic Raytracer

Jackson Kruger

Normally, the source code would be on Github, but because of some issues that popped up last-minute, here's a zip of the code used for the images seen here currently.
For this assignment, unlike my previous two, I did this development on Windows in Visual Studio.

Currently functional features (which should be apparent form the outputs below)

• Camera placement, film resolution, aspect ratio
• User specified background colors
• BMP output
• Spheres
• Ambient lights, point lights, shadows, multiple lights
• Basic sampling
• Diffuse, specular, ambient, and (recursive) reflective lighting

Architecture Overview

The file containing main, Raytracer.cpp, only exists to kick things off. It tells Parser what file to parse, who then initializes the relevant objects and sets appropriate values. The most important objects it initializes are the Renderer, the Scene, and the Camera. The Scene contains all of the lights and shapes (SceneObjects), while the camera primarily exists to construct the rays. The Renderer is the bridge between the Camera and the Scene, as well as the final Image object that is used to output at the end. It iterates over width and height, telling the camera to construct the rays, getting the colors of the pixels from the Scene, then setting the pixels of the Image. The Scene iterates through its list of pointers to SceneObjects, telling each sceneobject to intersect with the current Ray, producing an Intersection. Intersections currently are just a struct of Vector3s and a Material, but will soon be objects in their own right with a pointer to the Ray and SceneObject who produced it, as well as a Vector3 for the normal.

Challenges So Far

• The direction for the camera defined in the scene file confused me, particularly when it was posted that this direction is actually from the viewing plane to the camera. My raycasting was working imagining the direction as being from the camera to the plane, but it was horizontally flipped. I tried for quite a while to find a mistake in my ray generation code to accomodate literally flipping the direction vector on parsing, but that simply turned the camera around. In the end I just fixed the horizontal flippage with a little hack, multiplying by -1 in the right place.
• Refraction has been architecturally painful so far, as I've prepared to implement it.
• The exact meaning of max_depth wasn't clear initially, whether it meant that many reflections (rays + 1) or that many rays total. I wasted some time debugging my recursive reflections before realizing that I was interpreting max_depth as the max number of rays to generate total, not the max number of bounces.

Output Images

Sample scenes

The following are the rendered outputs of the provided sample scenes. The only modifications I made to these before running them were commenting out the output_image line as I haven't yet bothered to parse desired output_image location. I currently just output in the working directory to a file called "raytraced.bmp".

Custom scenes

Each of these is just a sample scene with another point_light added.

Resolution and Aspect Ratio

These are just to show that my Raytracer can handle odd FOV's and resolutions (they have hover text)