LOGAN SMITH
COmbat/technical Designer
BeAST of the West
Platform | PC Engine | Unreal Engine 5 Duration | In Progress Team Size | 3
Playtime | 30 min Role | Combat Designer
Responsibilities ​​
​​
-
Desgining and Implementing Gameplay Systems
-
Utilized CommonUI for UI/UX for the player character and gamplay loop
-
Designed and Implemented Enemy AI Logic and Actions with a Behavior Tree & State Tree Blend
-
Itterated on the 3 C's for the Player Character
-
Worked on a Combat Camera for 1v1 engagments for a more cinematic feel most notablity seen in the boss battle
-
Updated/Modified Multiple Traversal Systems to more scabable and reusable throughout the project
-
Added a Stealth System and Parry/Poise System to add more depth to combat.
​
Ghost Inspired AI and Stealth System
Design Philosophy
Inspired by Ghost of Yotei and Ghost of Tsushima, the goal of this AI system is to create enemies that are challenging, predictable, and fair. The system is designed to give players readable openings to master their environment.
Under the hood, this is achieved by blending the performance of a custom C++ State Machine with the flexibility of Unreal Engine 5's AI Perception and Behavior Trees.
Core Perception Mechanics
Standard Line of Sight (LOS) often feels binary and unfair in stealth games. To fix this, I added custom perception rules focused on the player experience:
-
Vision Zones (Forgiving Margins): Suspicion doesn't fill at a flat rate. If a player is caught dead-center in an enemy's vision, they are spotted twice as fast. If they are caught in the periphery, suspicion fills at half speed. This gives players a crucial split-second to realize their mistake and duck back into cover.
-
Vertical Blind Spots: To support vertical level design like rooftop assassinations, a custom height-limiter prevents enemies from unrealistically spotting players directly above or below them.
-
Responsive Hiding: A common bug in stealth games is AI staying locked onto a player who just jumped into cover. This system rather than checking when perception is updated it check against what actor it currently percieves leading to situations where if a player drops into tall grass while being watched, the AI instantly breaks visual lock and investigates the last known location instead of cheating.
