This project represents my first attempt in learning reasonml and writing a raytracer that uses bouncing rays. It was done both as a learning exercise for myself and as a showcase of my programming skills for Coursebase.
Basically a collection of rays from a single point are extruded in a pyramid shape. A 2D grid intersects this pyramid of rays where each grid square represents a pixel in a picture and each ray that intersects a pixel represents a ray of light.
The color of each pixel is determined by whether or not a ray that intersects the pixel also intersects with an object or the background color.
I've had experience writing a ray tracer before but this was my first attempt at writing a raytacer that simulates bouncing light. The simplest ray tracer usually just approximates light using a simple equation called Phong to calculate the color of an object based off the angle between a surface normal and one or more point sources of light.
This ray tracer is a bit more advanced then a phong ray tracer as it simulates real light using a recursive ray bouncing algorithm. The light source is also more complex in the sense that it is no longer a point light source but the entire sky serves as a source of light. Every point on an object has rays of light from two main sources: rays of light from the sky and dimmer rays of light bouncing off of other objects.
The ray tracer samples a random set of these rays hitting this point and averages the color to determine the best color to be drawn on the pixel. To generate the shading in the picture the more times a ray has bounced the darker the color. This results in shadows and shading that are very realistic as it simulates how real ambient light works.
The ray tracer is actually a monte carlo simulation that uses randomness to simulate light. The randomness can produce artifacts that appear as graininess in the picture... the way to reduce graininess is to increase the sample size of random rays but that comes at the cost of rendering time. In reality an infinite amount of rays hits every point on an object and the average brightness of all those rays produces a color.
There's no cheap way to simulate infinite rays or even fetch the average color of all those rays so the only realistic way to approximate the color is to pick a random set of rays. These rays are generated by going backwards. Rather then drawing a random ray from a light source to object then bouncing that ray to the camera, the whole random generation is done in reverse. The ray is drawn from the camera and bounced off the object all around the space until it hits the sky (aka lightsource).
The picture included in the repo: 800x400.ppm took hours to render with 200 random ray samples per pixel. The result is shown below:
Current settings in the file will take 5 minutes to render the same picture in the repo at 200x100 pixels with 100 random ray samples per pixel.
You can view the included picture by opening 800x400.ppm file in your favorite picture viewer. Some applications can't view the ppm but the mac previewer should work. If you can't view the ppm file I've included a 800x400.png file that's more compatible with all viewers.
After building there should be a file called Main.bs.js in the source folder. You can redirect stdout to a file to generate the picture.
node src/Main.bs.js > picture.ppmyarn clean
yarn build
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent