Back

Tool & AI

Tool

Unreal Engine 4 / 5

  • Blockout level production
  • Blueprint-based event and object design
  • Level Sequence
  • Animation Montage editing

Photoshop

  • Image editing

AI

AI Tools Used

Planning Support

Gemini, Claude

Image Reference

Nano Banana, Midjourney

Tool / Prototype

Antigravity

HTML-based tool & mini game production, UE5 integration

AI Usage Principles

I use AI not as a simple Q&A tool, but as a supporter that co-operates through the planning-development-evaluation cycle.

For most tasks, I follow this workflow:

Define Purpose

Summarize the project goal and desired outcome concisely, then expand step by step

Feasibility Validation

Confirm what can be built, and discuss with AI whether to start full production or a prototype first

Pipeline Setup

Assign roles (Planner / Operator / Evaluator) to AI and define role-based workflows. Establish project guidelines, folder structures, and naming conventions collaboratively, designing a reusable structure portable to other projects

Use Cases

HTML Level Prototype

Co-authored an 8-floor vertical dungeon level design document (floor-by-floor theme, gimmicks, flow, combat & puzzle design) with AI

Implemented a Three.js-based 3D blockout tool via AI, featuring data-driven floor auto-generation, multi-view, asset placement, and coordinate export

Built an environment to pre-validate spatial feel, sightlines, and vertical structure in the browser before engine integration

HTML Mini Game

Ideated a lockpick-style mini game inspired by Fallout and Elder Scrolls series with AI, then built it as an HTML prototype

Used to pre-test and validate the fun factor of mini game and puzzle elements before engine implementation

PaceMapper — Level Pacing Pre-Validation Tool

A pre-production tool enabling quick pacing balance review of monster spawn and item placement on a 2D map during the level planning phase

Built to reduce iteration costs in the engine and communication overhead with other teams — enabling simulation before engine work and communication based on visualized output

Automated object placement using Poisson Disk Sampling algorithm

Automatically calculates and suggests optimal placement intervals based on player movement speed and discovery frequency inputs

UE5 Vibe Coding

Set up a vibe coding environment by integrating Unreal Engine workspace with Google Antigravity

Used Antigravity to gain hands-on experience with C++-based level object creation, UMG setup, Animation Notify additions, and various prototype work

Level Design Quantitative Analysis

Conducted AI-assisted R&D to quantitatively analyze spatial density and exploration rhythm in open-world level design.

Designed the analysis framework and validation criteria personally, delegating large-scale data processing and script implementation to AI — a "planning-led, AI-executed" structure

Discovered design patterns and common principles across games from the analysis results, then derived insights through AI discussion based on personal planning intent and experience

Documented the methodology and results as a reusable framework asset ready for immediate use in future analyses