Dreamville Mart: AI Multiplayer Game
Owned the end-to-end UX/UI across onboarding, item input, in-game guidance, 2D progress feedback, and final states.
Overview
Highlights
Completed Playable Build
Delivered the full player-facing UX/UI and implemented the final interface in Unreal Engine.
CMU ETC Festival Showcase
Presented as a playable experience at the CMU ETC Festival.
Concept
A two-player desktop game concept developed by the Art Direction Lead, where each player creates three hidden items for their opponent to find, with AI-generated riddles available as optional clues.
Players
Input 3 Items for Your Opponent
AI System
Generate
Items & Riddles
Players
Explore the
Store
Players
Use Optional Riddles
Players
Find All Opponent’s Items
Game System
Show Winner
& Reveal Items
Challenge
Making a New AI Game Loop Easy to Learn
Guide players through item creation, AI-generated riddles, and exploration without making the experience feel complicated.
Keeping Players Oriented During Gameplay
Provide clear progress, status, and contextual guidance while keeping the interface lightweight inside the 3D world.
Overall Process
Understand
Define
Design
Implementation
Playtest
Iterate
Delivery
Design
My Approach
Structure the Player Journey
Organized onboarding, item input, gameplay, and final states into a clear sequence.
Use In-World Feedback
Integrated prompts, status updates, and clues into the game world to reduce UI friction.
Design for AI Variability
Created flexible UI patterns for unpredictable text input and variable AI-generated riddle lengths.
Experience Evaluation
Playtest & Iteration
Refined the functional build through weekly usability tests, then validated the experience with public players during two open-house testing days.
View weekly build walkthroughs → CMU ETC project archive
FINAL DESIGN
Experience Walkthrough
01 · Player Item Input
Structured three AI-driven item inputs into a clear step-by-step experience with visible progress and input guidance.
02 · In-Game Guidance
Combined clues, progress tracking, status updates, and a persistent shopping list into a lightweight in-game UI.
03 · Final Reveal
Revealed the winner alongside each player’s hidden generated items, giving both players a clear sense of what was found and what remained in the scene.
IMPACT
Project Outcomes
Generative AI as Core Gameplay
Used ChatGPT 3.5 and DALL·E 2 to generate riddles and player-created store items in real time.
Completed Playable Build
Delivered the full player-facing UX/UI and implemented the final interface in Unreal Engine.
CMU ETC Festival Showcase
Presented the completed playable experience at the Carnegie Mellon University ETC Festival.
