Roomba Rumble

7DFPS Game Jam, 2020 | See it here

For the 7DFPS game jam in 2020, I developed a multiplayer shooter game where players control weaponised cleaning robots.

For playability's sake, a concession had to be made and technically the camera placement is more similar to a third person perspective.

I've decided to continue with the game's development after the game jam had ended. As a result, I believe I have learned quite a lot about all facets of game development, from design, programming, time management, and so on. While the game was clearly far from a technical accomplishment, it served as a good introduction to learning gameplay programming, networking (using Photon initially, and porting to Unity MLAPI later on), shaders (using Shader Graph), as well as generally the Unity engine itself.

Unity

Blender

Driving physics

Self learning, 2022 | See it here

I've always found vehicle physics in video games to be an interesting problem, which prompted me to make small experiments in my free time trying to implement my own solution.

Having studied the well-known Marco Monster paper on car physics, I began toying with an implementation using the Godot engine. Within a few weeks I had a basic demo with a simple weight transfer that responds to changes in acceleration, raycast-driven wheel suspension, cornering that adapts to velocity and tyre friction.

It is by no means a complete solution, but it's been eye-opening in just how complex this topic - which many outsiders consider trivial - actually is. I'm planning on revisiting it at some point again; I believe the main issue I've faced is an over reliance on built-in rigidbody physics and collision detection; what I've learned is that vehicle physics, especially for wheeled vehicles such as cars, mandate that all interactions with the physics world are performed using bespoke mechanisms and approximations instead of the main rigidbody solver and triangle geometry (e.g. how raycasts and distance functions offer a smoother collision response for wheels than using the general physics hit event).

Godot Engine

Blender

Survive the Hunt

Video game modding, 2021-present | See it here

Survive the Hunt is a custom gamemode I've built in FiveM, the community-made multiplayer client for Grand Theft Auto V.

The idea came from watching the YouTube channel "FailRace", who popularised this mode while playing GTA Online on console. However, I felt that this would be difficult for ordinary gamers to enjoy without a coordinated group, as there was no guarantee players would follow the rules (whereas the FailRace group were tight-knit and therefore could agree on fair play for content creation purposes). Hence, the goal was to implement the ruleset as a FiveM script, adding the constraints and restrictions needed to guarantee a fair experience for all players.

At its core, the script implements the basic ruleset: a "hunted" player is picked at random. They're hidden from the hunters' radar and given a one minute headstart to establish an initial position and strategy. During that prep phase, "hunters" are prevented from leaving the starting area. After that, the hunted player needs to survive 24 minutes (12 in-game hours) in the city while their approximate location is broadcast to the hunters every minute.

C#

Survive the Hunt

Video game modding, 2021-present | See it here

I don't necessarily consider this project critical to my career in any way, and it was born more out of my own interest in trying the gamemode with my friends, but despite that I still found I gained some knowledge and skill from developing it over the years.

For one, working on the project I had the opportunity to delve deeper into the game design side of things, specifically balancing in a competitive multiplayer context. It's not something I get to do a lot, and it's not something I see myself working in, but I feel like it provided me with a perspective on how to gauge player satisfaction, how feedback affects future iterations of a product, timescales, the QA process, and so on. In fact, in some areas I began to slightly deviate from the original concept and added or adjusted features/rules as necessary based on playtesting feedback.

From the technical side of things, most of the code boils down to calling functions from the game's executable in my C# code in order to update the player, world or HUD state as needed, so it was more analogous to tasks performed by mission scripters (with the added caveat of having to interface with the game from external code).

C#

LepusEngine

Self learning, 2023-present | See it here

This is a toy 3D engine I've been writing in my spare time for the purposes of learning C++ and graphics programming.

I've initially started tinkering with OpenGL around 2017 and got a basic OpenGL renderer working, but unfortunately it suffered from tech debt due to my lacking C++ knowledge. In 2023 I revisited the project and decided to rewrite it from scratch.

Currently, the engine can render a demo app with basic animated primitives, and it can use either of the two backends (OpenGL or Vulkan) with only minimal dependencies. I've been using this project as an opportunity to brush up on my understanding of low-level programming and architectures.

I try to limit the project's reliance on the standard library and math helpers to a minimum, implementing utility code on my own until I feel that it's becoming unfeasible.

By far the biggest challenge in the project was getting to grips with resource management, synchronisation and overall API design in Vulkan, but graphics programming has always been an area I wanted to explore, so learning about those topics bit by bit has been fascinating.

C/C++

Vulkan

OpenGL & GLSL