Untitled Wizard Game

I wanted to make a spellcasting game from scratch in OpenGL.

What I got working:

• Rendering — forward Blinn-Phong with shadow mapping and an HDR → bloom pipeline.
• Physics — rigid-body AABB dynamics with impulse resolution, restitution and friction.
• World — infinite noise terrain, a Gerstner-wave ocean, procedural sky, and instanced wind-animated grass.
• Effects — a GPU-instanced particle system, and a castable refracting water vortex that lifts and traps whatever it catches.
• Content — hot-reloadable, data-driven .level files and spell definitions.

By the numbers. No engine, no physics library, no renderer library — the renderer, rigid-body solver, particle system, terrain generator, shadow mapping, HDR/bloom pipeline and level loader are all first-party. Measured in a --bench mode (fixed timestep, scripted camera path, per-pass GL_TIME_ELAPSED timer queries), median of three runs at 1080p, vsync off, release build, on an integrated Radeon 780M:

• 188 FPS mean / 137 FPS 1% low (5.31 ms/frame, stddev 1.24 ms), with the 2048² shadow pass at 0.469 ms.
• ~450,000 live particles and 2,560 rigid bodies each sustained above 60 FPS.
• Instancing versus an identical naive draw path: 4,149× fewer draw calls and 3.9× faster frames at 50K particles.
• 1,000,000 instanced grass blades (9.1M triangles) in one draw call at 79 FPS mean.
• 740 ms cold start; 5.10 MB static Windows binary (MinGW cross-compile).

However, since I started the project after having spent only a few weeks of skimming LearnOpenGL and with some carry-over experience from a computer graphics course in college, the project became unwieldy relatively quickly.

As such, I’ll re-implement the current progress off of better game engine design principles that will hopefully make it more easy to work with. Furthermore, when I was working on it earlier, I was more focused on going fast and neglected to maintain a proper devlog, so this redesign will also be a good opportunity to have a good devlog.

Details of the state of the project before this reimplementation is in the Wrap-up post.