Studycase: Hack and Slash prototype with Unreal Engine

if any image is not shown, please open it in a new tab

PROJECT LINK

Project link

Release note

DEMO VIDEO

TECH STACK

⚙️ TECHNICAL INFORMATION

• Engine: Unreal Engine 5.4.4
• Source: Blueprint & C++
• Demo video: Here or in the download panel below.

🎮 HOW TO PLAY

💡 Try mixing light and heavy attacks with directional dodges to explore different combos and behavior reactions.

⚙️ FEATURE DECISIONS

🔗 COMBO SYSTEM

While not the core focus of the current test, the combo system is critical to gameplay. It defines how both * players and AIs perform actions*, and it decides the flexibility allows for dynamic, scalable design.

✅ Design Goals

• Scalable and designer-friendly
• Clear separation between combo animation and logic tree
• Lower dependencies between design combo flows and present combo action

After extensive iteration, the final structure includes the following components:

🧩 Components

ComboInput

• Represents basic input actions (Light Attack, Heavy Attack, Counter, Dodge, etc.)
• Used to traverse the combo tree

ComboTree

• A Trie-like data structure for navigating input sequences
• Starts from a root (default state) and moves through valid input chains

ComboSubsystem

• Central system for managing combo trees
• Builds and manages combo trees for all relevant actors

ComboRunner

• Manages player interaction with combo trees
• Fetches tree from ComboSubsystem, progresses based on input
• Validates input at each stage to determine the next combo node

ComboAnimationPlayer

• Executes animations based on current combo state
• Communicates with the ComboRunner to play appropriate animations
• Uses Animation Montages with Combo Window Notify States to control combo timing windows
• (WIP) use “Combo Fastfoward State” to increase combo speed and smooth

📊 Diagrams

🗃️ Data Asset views

🤖 AI BEHAVIOR

Thanks to Unreal Engine’s native support for:

• Behavior Trees
• Blackboards
• BT Tasks and BT Conditions

AI logic is cleanly implemented and developer-efficient.

AI Perception

AI can detect the player through:

• Sight
• Sound
• Custom scripted triggers

🧭 AI Modes

⚔️ AI can execute combos, such as Player do because they send the same input flow as player’s.

⚔️ To briefly demonstrate fighting impacts, hit actions now PUSH targets away instead of causing damages.

🟡 Wander Mode

• AI walks between 4 predefined waypoints
• Pauses at each waypoint for ~8 seconds
• Randomly plays idle/fun animations
• When disengaging from combat, AI returns to the nearest waypoint

🔴 Engage Mode

• AI approaches the player when in range
• Stops at appropriate combat distance
• Behavior includes:
  • Evading player attacks
  • Counter-attacking upon successful evade
  • Flanking if behind the player

🌲 Behaviour Tree

🕺 PLAYER ANIMATION

• Base animations reused from Unreal Engine Mannequin
• Customizations added for immersion and responsiveness

Enhancements:

• Randomized idle animations to add realism
• Player automatically draws sword when:
  • Enemy is near
  • Enemy is visible via the camera system
• AI automatically draws sword when see or hear Player (via AI Senses).