Engine for skill demonstration purposes assembled without tutorials.
- For GLFW3 reference their documentation.
- For OpenGL reference their documentation
- GLAD vs GLEW on stackoverflow. I generated GLAD on this website. I generated all because I didn't feel the need to read through what do I need as I only wanted to get started and I wasn't under restrictions.
For build instructions see vcpkg.md because it's package manager so I figured building documentation should be there.
Babylon documentation - First Scene and Model
Whether you are creating a whole world or just placing one model into a web page you need a scene to contain the world or model, a camera to view it, a light to illuminate it and, of course, at least one viewable object as a model. All models, whether just a box or a complex character, are made from a mesh of triangles or facets.
- Scene
- Camera
- Light
- Object
Reading Babylon.js implementation
To understand engine concepts, I read implementation of Babylon.js's scene, engine, simpleEngine, abstractScene, material, mesh, shaders.
Locations:
abstractSceneenginestandardMaterialscene- Smallest possible API
- Documentation to see the class relationships
I talked with ChatGPT to create conceptual map of how to implement the rendering API. After roughly 16 hours and an attempt to design public api using structure below, I took a walk because I felt burnout. While walking I thought about different things and how to complete renderer in time.
I realized my weakness was that I don't have great grasp of how to use OpenGL API so I decided to put together code with help of ChatGPT starting from concepts going to the code to gain more complete understanding. A solution was to create something fairly complex: A movable camera in 3D world with help of glm because I have yet to hone mathematical skills to do the math myself. 3D objects in the world that I can observe as I move the camera using mouse and keyboard controls.
That would yield me some mental images of how it fits and works together, how to use OpenGL API and give me ideas of most frequently called commands and how to assemble my own API that would allow me to add meshes to the scene in somewhat user-friendly way.
With those fundamentals I could approach conceptual design again and hopefully make it work. I disliked the feeling that I don't understand every single line of code because I had fear to change stuff because I didn't know if it will break something. However, I could make all of that work together, so I guess my conceptual understanding was solid because I knew what logic I need and where to place it. Obviously, code wouldn't work if I didn't know how it works... duh!
After digging through the Babylon.js engine and attempts to figure out all the math required to setup camera, I feel like it is a lot... In their engine I saw that they have at least three cameras and they involve matrix computations. Matrices on their own are OK, while I am not that great with trigonometry because I need practice. I learned how much math the engine requires because all data that goes through the engine relies on mathematical computations. Not even the glm library helps with math if I don't have in my mind how it fits together. Once one matrix fails and doesn't do what I expect it to, that's it, I don't know which value is messing it up.
OpenGL and learning that on it's own is fine, however there is a lot math going on which I should focus and learn to effectively work with computer graphics. Currently it seems primary targets are linear algebra and trigonometry, however I believe there are other areas of math that play crucial role. Perhaps algorithms for object surfaces.
I am confident I could create a Node class that contains object data and takes in a texture or whatever which is just an OpenGL program which accepts uniforms and I could play with different matrices and objects in the scene that way... I could have some kind of API doing that! The current code that creates a window and renders data to it could be a separate thing for itself as a documentation of the process so far... I had so much fun doing all of that! I feel like I could just keep going, and damn I really need to learn the math.
I've read implementation of various methods and classes of Babylon.js to gain understanding in some common patterns that widely adopted engine uses. I've read so much and I feel like I have so much information with very limited experience so far. They use advanced methods that I am not familiar with, however I've gained quite a few ideas on what to do now. I could play with frustum, matrices, rotations, etc... Figure out math by pen, paper, tutorials, books, etc... then implement that math using OpenGL and programming!
Let's let it sink in!
I find the project extremely helpful because I can use it to build new knowledge on the previous tinkering. I don't have to set up everything from scratch again, mess with library setup, repository setup, etc... I can document references and find them later. I find the idea of doing it like this very helpful. I wouldn't have gotten here if I didn't start working on sk-experiments in march 2022. I am glad I did.