Lumina Studio
AI Music Platform — Create · Play · Earn
Create music videos in minutes · play to learn the music · earn in the marketplace
SELVAKUMAR Sachin 250941Z · Immanuel Khoo 253680U · Kiu Fa An 254400K
Client: Violet Tay (RememberOurStoriesInSG) | Module: IT2114 Digital User Experience Design | Assignment 1
Presentation Overview
Executive Summary
One platform, three connected pillars — Create, Play & Earn
Lumina Studio, built for client Violet Tay (RememberOurStoriesInSG), is an AI music platform that turns one idea into a finished, shareable music video — then turns that song into a game people play, learn from, and pay for.
Create — the Music Video Creator collapses a 40+ hour manual production (lyrics, composition, filming, editing) into minutes, via Gemini 2.5 (lyrics) → Lyria (music) → ComfyUI (video rendering).
Play & Learn — every song becomes a Rhythm and Piano game that gamifies learning its beats, notes and melody. Earn & Share — the Marketplace lets creators publish and players buy songs, powered by a leaderboard-and-energy economy.
Create · Play · Earn
A creator-economy music platform: AI-powered music video generation + gamified play-to-earn marketplace.
The Music Video Creator makes finished music videos effortlessly — a 40+ hour manual pipeline done in minutes. Type an idea; AI writes lyrics, composes the song, and renders cinematic scenes.
The Rhythm & Piano games turn each song into engaging play — players earn AI credits by hitting notes and climbing leaderboards, unlocking songs and generation credits.
The Marketplace is a self-sustaining creator economy — browse, play-to-buy, leaderboard multipliers, an ad-funded energy loop, and a creator revenue share.
Platform Overview
Key Features
Turn any idea into a catchy, original song with AI lyrics, melody and professional vocals — no music skills required.
Auto-generate cinematic music videos from the song — stitched from AI scenes, rendered and exported in HD.
A falling-note rhythm game built from any song — learn its beats and timing by playing, scored with combos & accuracy.
Virtual piano with a falling-note guide that teaches a song note-by-note — adjustable density from chill to chord-chaos.
Browse and preview songs, play-to-buy, climb leaderboards, and earn points — creators publish and take a revenue share.
An ad-funded energy loop keeps play free: watch an ad to restore energy, convert points to AI credits, buy songs with IWSC watermark.
Technology Stack
Frontend
AI + Cloud Backend
Our Vision
Vision
To make music creation effortless, learning playful, and creativity rewarding — one platform where anyone can create a song, play it to learn it, and earn from it.
Mission
Give creators professional music videos in minutes, give players a fun way to learn music by playing it, and give the community a fair economy to share and monetize songs.
Core Values
Client Overview
| Challenge | Current Manual Process | AI Solution |
|---|---|---|
| Song Lyrics | Composer manually writes all lyrics from scratch for every project | Gemini 2.5 Flash auto-generates lyrics from a topic |
| Melody Score | Composer manually composes melody scores using traditional methods | Lyria AI generates full musical compositions in minutes |
| Music Production | Manual Logic Pro X arrangement & mastering — time-intensive | AI handles arrangement, mixing & mastering automatically |
| On-Site Filming | Travelling on-site to film scenery — costly and slow | Veo 3 / ComfyUI generate cinematic scenes from text prompts |
Workflow Transformation
Requirements Analysis
The Problem
Creating, learning and monetizing music are all slow, hard and disconnected — one problem per pillar.
Writing lyrics, composing, mixing in a DAW, then filming and editing — a solo creator burns days on a single video.
Watching a video isn't practising. Without active feedback, beats, notes and melody don't stick.
Songs scatter across YouTube and various platforms with unclear earnings — distribution friction stalls income.
Composition, mixing and mastering in tools like Logic or GarageBand have a steep curve most people never climb.
Our Solution
Why Lumina Studio?
Group Component
User Research · Synthesis · Information Architecture · Service Design
SELVAKUMAR Sachin · Immanuel Khoo · Kiu Fa An
User Research
Our research is grounded in 4 in-depth interviews — one with our client/stakeholder (Violet Tay) and three with target users across the create, play and earn pillars — supported by hands-on competitive review of existing AI-music and rhythm-game tools.
Interview Overview
Interview Template
| No. | Question |
|---|---|
| Q1 | Can you walk us through your current end-to-end production workflow for a single music video? |
| Q2 | Which parts of that workflow consume the most time, money, or effort today? |
| Q3 | How are your existing songs (lyrics, melodies) currently created, registered, and reused? |
| Q4 | What is your vision for using AI in this process — what would "success" look like? |
| Q5 | Who are the audiences and platforms you most want to reach (e.g. YouTube, the public)? |
| Q6 | How important is staying true to your original lyrics & melody when AI generates the video? |
| Q7 | What gamified or interactive experiences do you envision for your audience? |
| Q8 | What concerns do you have about AI-generated avatars, voices, or visuals? |
| Q9 | Which features are essential vs. "nice-to-have" within a student-project scope and budget? |
| Q10 | How will you measure whether the platform genuinely reduces your workload? |
Interview Template
Darren Wong Wei Jie
Aspiring Music-Video Creator, 32 · Side Hustler
Marketing professional who wants to build a music / music-video channel for passive income, but has zero music or video-editing background.
"I want to put music videos on YouTube, but I have zero music background. If an AI can do the heavy lifting, I'm willing to pay."
Rachel Lim Su Yin
Casual Player & Parent, 32 · Tampines
"My daughter remembers a song after one listen. If she can actually play it and learn the notes, that's so much better than just watching."
Mr. Farid Rahman
Institutional Adopter, 41 · Brings tools to a group
"My learners actually remember a melody they've played. If a platform lets them learn music by playing — and I can license it simply for a group — that's transformative."
Key Insights
Affinity Mapping
7 key themes clustered from user & stakeholder interviews
Insight Statements
Sub How-Might-We
How might we let someone with no music or editing skills generate a complete music video in minutes instead of 40+ hours?
How might we turn any song into play that teaches its beats & notes, so learning happens through the game rather than passive watching?
How might we let players & creators earn and access songs through a fair, free-to-play marketplace funded by points, ads & a revenue share?
Relevant Research to Support
Design Principles
Every feature should be immediately understandable. If it needs explanation, redesign it.
Smooth animations and micro-interactions make the experience feel polished and engaging.
Don't lecture — let people learn a song's beats & notes by playing it in the rhythm & piano games.
Show only what's needed at each step. Advanced options exist but never overwhelm.
Every action gets a visible, immediate response to maintain engagement.
User Personas
Based on our research, we identified three primary personas representing our target audience. Each has distinct needs, goals and pain points that shaped our design decisions — and each is owned by a team member in the Individual Component.
Marcus Tan
Content Creator & Side Hustler
Marcus is a content creator juggling a day job with a YouTube music channel. He spends 40+ hours per video on lyrics, composition, filming, and editing — time he could spend creating more content or earning. He needs AI to handle production so he can focus on creative direction and publishing.
Rachel Lim
Casual Player & Parent
Rachel is a 34-year-old parent who wants her kids to learn music in a fun, interactive way — not passively watching videos. She plays music herself and believes active engagement (like learning piano or rhythm) sticks better than passive listening.
Mr. Farid Rahman
Institutional Adopter / Group Buyer
Mr. Farid Rahman is a music enthusiast and community organizer who wants to introduce engaging music experiences to groups — whether friends, family, or community members. He's looking for a platform that's easy to share, fun to play, and doesn't require musical expertise.
Persona Comparison
| Aspect | Marcus Tan | Rachel Lim | Mr. Farid Rahman |
|---|---|---|---|
| Primary Goal | Fast, monetizable music videos | Fun way to learn music by playing | Engaging, safe platform to adopt |
| Key Metric | Time-to-publish | Songs played & learned | Group engagement |
| Pain Point | Tool fragmentation, copyright | Passive apps, paywalls | Passive tools, group setup |
| Platform Use | Creator Studio | Music Game (Rhythm + Piano) | Marketplace + Group Admin |
| Success Measure | Videos shipped & shared | Songs her kids play unprompted | Rolled out to a whole group |
| Owned By | Sachin | Kiu Fa An | Immanuel Khoo |
Information Architecture
Marketplace is included as the ecosystem bridge connecting Creator Studio, Music Game, Learning Hub and group adoption.
User Flow
The flow now connects Sachin's Creator Studio, Immanuel's Marketplace bridge, and Fa An's Music Game into one repeatable ecosystem loop.
Service Blueprint · Creator
Service Blueprint · Learner
Combined Service Blueprint
Individual Component
Persona · Journey Map · Wireframes · Hi-Fi Prototype & Design Rationale
SELVAKUMAR Sachin
Creator Studio & AI Pipeline · Persona: Marcus Tan (Content Creator)
Persona · Marcus Tan
User Journey Map · Before App
Goal: Create and monetise SG-themed music videos manually while juggling limited time, tools and skills
User Journey Map · After App
Goal: Build $300–1000/month passive income from SG-themed AI music
Creator Studio · Wireframes
Hi-Fi Prototype & Design Rationale
Use of AI Tools & Declaration
Create Flow — Topic to Music Video
Entry hub ("Compose, Create, Captivate") routing to Creator Studio or the Music Game. States: empty (no productions yet) and populated — production stats, How-It-Works, Studios & Modules grid, live render progress, and a library of finished songs.
3-step AI wizard. Step 1 Lyrics: paste/write → "Enhance with AI" chat asks clarifying questions → approved lyrics. Config: genre, BPM, key, vocal (Soprano AI + Choir), chord progression, scene list. Step 2 Render Pipeline: Narrative / Continuous / Text-to / Image-to-Cinematic + voice → Generate, then live progress.
Cloud render tracker (e.g. 2/4 scenes), scene-by-scene preview, then the full player with synced Lyrics · Composition · Scenes panels. Shows ComfyUI/WAN render status cards, retry, and the in-browser FFmpeg "Merging clips" step; empty state offers "Generate Video".
KIU Fa An
Music Game · Play & Learn · Persona: Rachel Lim (Casual Player & Parent)
Persona · Rachel Lim
User Journey Map · Before App
Goal: Help her child learn songs actively, safely and affordably without relying on passive music apps
User Journey Map · After App
Goal: Discover Lumina, learn songs by playing, earn points to unlock more — and get her kids hooked
Music Game Home · Wireframes
Hi-Fi Prototype & Design Rationale
Use of AI Tools & Declaration
Play Flow — Idol Music Academy
"Idol Music Academy" landing — three modes (Play Song, Create Song, Piano Mode) plus feature cards: Rhythm Game, Piano Practice, AI Song Creation, Score & Progress.
Song picker with difficulty filter, custom-URL / file upload, and star-rated song cards. Two views: Rhythm mode and Piano mode; custom-URL-ready state validates the track before play.
4-lane falling-note gameplay. States: difficulty select → "Analysing your track" beat-detection → playing (Score / Combo / Accuracy HUD) → AutoPlay demo → Paused overlay.
Virtual keyboard with a falling-note guide. Note-density selector (2-bit Chill → Mix Chord-chaos); playing state shows falling notes, Score / Streak / Hits, and an upcoming-notes queue.
Immanuel Khoo
Marketplace & Monetization Engine · Persona: Mr. Farid Rahman (Institutional Adopter)
Persona · Mr. Farid Rahman
User Journey Map · Before App
Goal: Find a safe, engaging music-learning tool that can work for a group, not just one person
User Journey Map · After App
Goal: From discovering the platform to rolling the play-to-learn marketplace out to a group
Learning Hub · Wireframes
Hi-Fi Prototype & Design Rationale
Use of AI Tools & Declaration
Earn Flow — Results, Leaderboard & Marketplace
Post-game results with an animated rank grade (S / A / B / F), final score, accuracy, max combo, star rating, and a hit breakdown (Perfect / Good / Miss). Actions: Play Again · Choose Song · Home. The grade and points feed the leaderboard multiplier.
The monetization loop in one flow:
Screen Gallery — All 9 Core Screens
Reverse-engineered low-fidelity wireframes covering every core screen of the working Lumina Studio build — structure & flow first, before hi-fi polish.
More Live Screens
Beyond the three core flows — these supporting modules are also live, interactive screens in the working build.
Colour System
Typography
Component Library
System Architecture
Song Generation Flow
Key Technical Features
Real-time beat detection, falling-note generation, hit detection with timing windows, combo system and smooth canvas rendering.
Virtual piano with sound synthesis, falling-note guide, note highlighting and adjustable note density.
Multi-step pipeline: lyric enhancement, music composition, video generation, FFmpeg merge and karaoke sync.
Compute orchestration, metadata, media storage and a marketplace / leaderboard data layer with streaming delivery.
Architecture Evolution
We outgrew our first setup and re-platformed onto AWS for heavy AI render & storage.
Security & Privacy
Product Roadmap
Success Metrics & Targets
Illustrative launch targets — the loop only works if engagement funds the compute.
Ecosystem & Distribution
We're not claiming corporate partnerships — these are the providers we build on and our routes to market.
Challenges & Mitigation
Use of AI Tools & Declaration
Key Takeaways
Thank You
Open for questions — let's discuss Lumina Studio
IT2114 Digital User Experience Design — Assignment 1