Movement System

Overview

The movement system was completely rebuilt to support flexible, responsive player control through a Hierarchical Finite State Machine (HFSM). Earlier iterations suffered from floaty jumps, velocity issues, highly limited state flexibility, and bad code practice. By starting from scratch and decoupling movement logic from legacy systems, the final system supports modular state behaviors, dynamic interactions, and smoother, more reliable movement.

Contributed Breakdowns

Guncaster Homepage
Level Design

MOVEMENT SYSTEMS

Audio System
AI Programming
Backend
Development History
Studio History
Gameplay Showcase
Project Pipeline

Features

Some of the features built in for gameplay enhancement:

  • Fully modular HFSM-based movement states

  • Movement tech like dash jumping and wall springing for advanced players

  • Animation curves for fine tuned controls

  • Wall Kicks to enable vertical recovery and range

  • Slope handling to improve terrain navigation

  • Decoupled, scalable structure shared with AI for consistency

Timeline

Iteration 1

The first iteration of the movement system featured a factory pattern. It’s states were built at runtime and almost all of it’s values lived in the script itself. After deliberation I decided to take over the movement system and refactor it in iteration 2.

Factory Pattern Finite State Machine

Iteration 2

After taking over I worked tirelessly to bring it to player and coding standards. First was shifting away from the clunky factory pattern and inheriting the Hierarchical Finite State Machine base. After the refactor it was responsive, well received, and despite limitations full of life and features.

Hierarchical Finite State Machine

A history

  • Discovery of Iteration 1

    During Guncaster's development, the movement system went through two major phases. At first, headed by a different team member, the movement system slowly developed through the first half of development, an already alarming amount of time. After I finished working out the AI state machine base I started looking into how I could match the state machines in the project. When looking into things it was a mess. Repetitious lines throughout the states, variables tucked into the code itself, unnecessary methods or overly referenced methods. After toiling on it for around 2 weeks I opted to rewrite the system and the team concurred.

  • Iteration 2 in production

    Starting nearly halfway through development the movement was finally being established, an oversight that me and the team had neglected to address. After inheriting from the base Hierarchical Finite State Machine that the AI units were using, establishing the basic state flow, and adding the movement math with exposed variables things were back up to speed. It took less than a week to make and the results were much better in both gameplay and for our designers. After getting done with the basics I aimed to bring the controller up to what our team had planned; wall running, dashing, crouching, sliding, and overall responsiveness.

  • Adding Spice

    I started with quality of life measures like making the movement snappy, the jumps less floaty, adding slope handling, then moved onto some features. Based on feedback I discerned that players didn’t like small environments and as such wouldn’t enjoy sliding under things in cramped spaces. Crouching and sliding were cut. Dashing was trivial and I opted for a velocity change over adding velocity to make level design easy and player movement predictable. Lastly was wall running, with little time left I decided to go for something similar but different. The result was wall kicking, essentially a double jump that can only happen near a wall. For extra spice I added features like accounting for wall normal and direction of the player for a fun and unique addition to the movement. The end result was a blast to play with and easy to change and debug with.

Showcase

What Went Right

The final design of the movement system feels really fun to play with, that in it of itself is an achievement I am proud of. Coming from where the movement was and where it ended at, I couldn’t be happier. The movement also has features that I think came out nicely and add an extra level to the mastery some players can achieve. This was slightly calculated but definitely originally conceived itself as a happy accident. The decision to let the airborne state affect air control and resistance differently let stuff like hopping to preserve forward motion really important. Combine that with the fact that dashing followed by immediately hopping would be just a bit faster than running on the ground do to the instant velocity change dashing offered. I also enjoyed the difference between a jump and wall jump. One was additive and the other was a velocity change. the result was in some instance upward motion combined and you were able to get a lot more height which was fun. I also really enjoyed how simple and how much wall kicking added to the game. It was definitely the additional layer the game needed to feel unique and fun.

What Went Wrong

Considering we left the movement system and it’s developer unchecked for so long was our fault collectively as a team. When it finally spiraled out of control we were forced to perform a rewrite which put a lot of pressure on the developers needing to test things and me the programmer scrambling to fix things when I had other systems to take care of.
Other than that an issue I wasn’t proud of was the fact that the slope handling works fine up hills but not down hills. When the player has enough forward momentum they can leave the bounds of the slope and go airborne before they land on the ground again. It felt almost like a staircase.
I also regret not being able to take full advantage of my features like wall kicking on oddly shaped surfaces to make more interesting levels. Of course I also had other feature ideas that didn’t make it which I was slightly bummed with. I had traps that could directionally launch the player, or explosions that would push the player outward like a rocket jump. Things that didn’t make it to the final build.