This readme has been adapted from the Documentation.txt file that I originally submitted with this project.
Fractal Exploration
by William Brigham, CAP4720 Fall 2019
Project Proposal
An exploration of different types of fractals and the ways that they can be rendered.
The goal will be to create examples of fractals from most, if not all of these categories: https://en.wikipedia.org/wiki/Fractal#Common_techniques_for_generating_fractals
I would like to demonstrate the computational intensity of rendering different types of fractals and hopefully be able to create some original fractals.
These would be implemented in GLSL fragment shaders, probably just being rendered on a simple quad mesh.
Ideally, I would just be using the Wikipedia pages for each fractal type as a resource. I may end up looking at some online papers if I need to and am able to find some.
Algorithms and Data Structures
All of the fractals shown are rendered onto a single plane mesh, meaning that all geometry is being constructed in real-time.
The 3D fractal demonstration uses the "raymarching" technique to draw the geometry. This means the use of signed distance functions and space-deformation functions in order to create the environment.
The Henon Map and Random Walk demonstrations use a double-buffering technique to maintain state. Basically, the result is rendered into two WebGLRenderTargets, which are flip-flopped between. This allows the previous frame's output to be fed into the shader as a uniform, creating a sort of feedback buffer. These are referred to as "compute shaders" within the source code of the project. In both of these demos, a single particle is being kept track of by using the bottom left corner pixel color as the particle's information (i.e. position, color, size, etc.). Everything else is updated by the particle's data, as well as being augmented by what was previously at the pixel. This allows the particle to effectively draw on the screen like a pen.
In the Henon Map demo, this particle is randomly re-spawned about every two seconds. This allows the space to get modeled in a better visual way without needing many different particles.
The quadtree demo could itself be considered a data-structure,
but no data is actually being stored in it, so /shrug.
Some distance functions used in this project are not ones that I came up with, such as the signed-distance formula for an equilateral triangle. There are comments in the source code of this project that link to where I got these functions from.
How to run this project
Just open the project.html page in a web browser.