Bachelor Project
Project Overview
Research Question
What are the specific advantages, disadvantages, and limitations of Unreal Engine 5 versus Unreal Editor for Fortnite when implementing AI behaviors for game characters, particularly for stealth-based gameplay mechanics?
Thesis Summary
My bachelor thesis investigated the capabilities and challenges of developing artificial intelligence for Non-Player Characters (NPCs) within two prominent Epic Games development environments: Unreal Engine 5 and the Unreal Editor for Fortnite. The research aimed to determine the efficiency, flexibility, and specific advantages of each platform for creating AI-driven game experiences, particularly in the context of a stealth game.
Methodology & Implementation
Research Approach
A key component of this research was the development of a stealth game prototype, "Starfall Relic," which was built in parallel in both UE5 and UEFN. This practical approach allowed for a direct comparison of AI implementation. The study employed the NGOMSL (Natural GOMS Language) model to systematically analyze and evaluate the development process.
Test Scenarios
The research involved three distinct tests of increasing complexity:
-
Test 1 - Basic NPC Navigation
Implementing a character moving from a start to a target point. -
Test 2 - Advanced Guard AI
Creating a guard NPC with patrolling behavior, player detection (sight and sound), combat engagement, and investigation of last known player positions. -
Test 3 - Populating a Settlement
Developing a diverse range of NPCs, including 30 guards (half on patrol), merchants, suppliers with delivery routes, general inhabitants, and key-holding guild members.
Evaluation Criteria
The evaluation process was comprehensive, analyzing both measurable and subjective aspects of the development experience:
-
Quantitative Metrics
Development time, number of steps required -
Qualitative Aspects
Ease of use, available tools, documentation quality -
Workflow Analysis
Testing efficiency, workflow clarity, technical limitations
Key Findings & Results
The comparative analysis revealed that neither engine is universally superior; the optimal choice depends heavily on the developer's specific needs, project scope, and desired level of customization.
Unreal Editor for Fortnite (UEFN)
Strengths
UEFN excels in rapid prototyping and development of simpler AI behaviors, largely due to its Creative Devices (e.g., Guard Spawner, Patrol Path Device) which offer pre-built functionalities. This significantly reduces setup time for common NPC tasks. It provides a more accessible entry point for developers leveraging existing Fortnite assets and systems, with a streamlined publishing process within the Fortnite ecosystem.
Weaknesses
For more complex or custom AI, UEFN necessitates the use of its proprietary scripting language, Verse. While powerful, Verse can present a steeper learning curve and increase development time, especially for those unfamiliar with it. The testing and iteration cycle in UEFN can become notably slower as project complexity and size increase, due to the need to launch a session and upload changes. Furthermore, UEFN operates within the Fortnite sandbox, which, while offering many assets, can limit absolute creative freedom compared to UE5. Direct interaction between custom NPC Verse behaviors and some Creative Devices required workarounds like a custom-scripted game manager.
Unreal Engine 5 (UE5)
Strengths
UE5 offers unparalleled flexibility and control for creating highly customized and complex AI systems. Its comprehensive suite of tools, including Blueprints and Behavior Trees (and C++ for deeper access), allows for intricate logic and fine-tuned NPC behaviors. The in-editor testing and iteration process is significantly faster and more direct. UE5 demonstrated strong performance in handling a large number of AI agents simultaneously, as shown in stress tests.
Weaknesses
The power of UE5 comes with a steeper learning curve and a more involved development process, requiring more steps and manual setup for AI controllers, character blueprints, and animation systems. While extensive documentation exists, finding specific, example-driven guidance for complex AI scenarios can sometimes be challenging. Assets often need to be created or sourced externally.
Conclusion
UEFN is highly effective for developers seeking to quickly create and publish experiences within the Fortnite ecosystem, particularly when leveraging its pre-built devices for standard AI behaviors. It can be considered an "Unreal Engine Mini," offering a good balance of power and accessibility for certain project types.
UE5 remains the choice for projects demanding deep AI customization, high visual fidelity, and complex game mechanics outside the Fortnite framework. It provides the tools for ambitious, standalone titles but requires a greater investment in learning and development time.
The research concluded that the "better" engine is contingent on the project's specific goals, the developer's experience, and the trade-offs between development speed, customization depth, and platform integration.