DDA FPS Game
A first-person shooter featuring dynamic difficulty adjustment using Large Language Models (LLMs) to adapt gameplay in real-time.
Watch DemoAbout the Project
Developed as part of a Bachelor Thesis, this FPS serves as a testing ground for Dynamic Difficulty Adjustment (DDA) powered by Large Language Models (LLMs). The system captures real-time performance data and utilizes an LLM backend to balance gameplay dynamically, ensuring a tailored challenge for any skill level.
Core Systems:
• LLM Data Pipeline: Tracks detailed round stats—including accuracy, headshots, health management, and ammo consumption—feeding them into the LLM for real-time difficulty balancing.
• Dynamic Spawner Logic: Enemy and resource spawns are triggered procedurally based on the DDA system's output to maintain optimal player "flow."
• Combat Mechanics: Features tactical sniper scope zooming with FOV reduction, smooth weapon switching, and a comprehensive health/medkit system.
• Zombie AI Behavior: Sophisticated enemy logic including patrolling, chasing, and varied attack patterns (melee and projectiles).
• Advanced HUD & UX: Includes a real-time minimap for situational awareness, a card collection system from fallen enemies, and immersive VFX/SFX (muzzle flashes, explosions).