Kapsule Lab
Built for the Cloud Run Hackathon 2025
Kapsule Lab was inspired by the frustration artists face spending more time making content and less time in the studio. It was built to speed up workflow, save time, and cut production costs, allowing artists to focus on what truly matters - making music.
Inspiration
As someone who works in the music industry, I see how much time artists waste creating social content instead of making music. Traditional tools required manual editing or complex AI prompts. Kapsule Lab removes those barriers - enabling creators to produce professional visuals and sound assets in minutes for a fraction of the cost.
What It Does
Kapsule Lab combines three powerful tools in one platform:
Vocal Lab – Generate professional-grade vocals using ElevenLabs Music API.
- Input lyrics or melody prompts
- Choose vocal type, mood, genre, and effects
- Export clean WAV or MP3 acapellas
- Input lyrics or melody prompts
Art Lab – Create track artwork using Vertex AI Imagen 4 and Gemini 2.5.
- Input a short text prompt describing your song
- Select genre, mood, and visual style
- Instantly generate 1:1 or 9:16 artwork ready for release
- Input a short text prompt describing your song
Video Lab – Generate short-form music video content optimised for social media.
- Upload a 15-second segment of your track
- Describe your desired visual scene or theme
- AI transforms it into cinematic, vertical clips powered by Veo 3.0
- Upload a 15-second segment of your track
What I Learned
- Prompt engineering drives results: Structured, detailed prompts deliver vastly superior video and artwork quality.
- Browser-side processing solves scale issues: Extracting audio segments client-side bypassed Cloud Run’s 32MB request limit.
- Async jobs improve UX: Firestore polling ensures smooth updates during 2–5 minute video renders.
- Aspect ratio matters: All outputs are optimised for 9:16 vertical video to suit TikTok, Reels, and Shorts.
- Gemini enhances creativity: AI-enhanced prompts yield richer, more cinematic scenes than static templates.
How I Built It
Prototyping
Started in Google AI Studio for early development and concept testing, then refined the UI and workflow in Cursor.
Frontend (Cloud Run)
Deployed as a static SPA using Tailwind CSS and vanilla JavaScript, hosted on Google Cloud Run under the kapsule-lab studio environment.
Backend (Cloud Run, FastAPI)
- Prompt preview and Gemini-enhanced refinement
- Veo video generation workflow
- GCS uploads for audio and video files
- FFmpeg processing for merge and looping
- Firestore job tracking for asynchronous operations
AI Models and Services
- Gemini 2.5 Flash: Prompt enhancement and creative text-to-video direction
- Imagen 3 (
imagegeneration@006): Artwork generation - Veo 3.0 (
veo-3.0-generate-001): Text-to-video rendering - ElevenLabs Music API: Vocal generation and audio synthesis
Infrastructure
- Storage: Google Cloud Storage (audio, video, temporary files)
- Processing: FFmpeg (audio-video merge and looping)
- Database/Jobs: Firestore (asynchronous status tracking)
- Payments & Credits: Supabase + Stripe for credit-based usage and payments
Challenges I Faced
- Handling large audio uploads within serverless request limits.
- Achieving consistent prompt-to-video quality across genres.
- Managing asynchronous Veo tasks and FFmpeg merging reliably.
- Ensuring CORS compatibility across multiple environments.
- Coordinating multiple AI tools and cloud services smoothly under one system.
Accomplishments I'm Proud Of
- Successfully deployed a fully functional serverless application using Google Cloud Run.
- Integrated Google’s AI capabilities — Veo, Gemini, and Imagen — into a live creative production workflow.
- Built a complete end-to-end system from file upload to AI generation and delivery.
- Overcame steep learning curves with Google Cloud Console, AI Studio, and Vertex AI integration for the first time.
- Early users from the Kapsule artist community are already testing and providing valuable feedback.
What's Next for Kapsule Lab
- Live Launch: Integrate directly into the main Kapsule app for artists to use alongside their music profiles.
- Monetisation: Introduce a credit or subscription system for premium features and faster rendering.
- AI Pipeline Improvements: Enhance model chaining for more consistent quality across media types.
- Creator Community: Showcase user-generated videos and artwork inside the Kapsule ecosystem.
- Scaling: Expand Cloud Run capacity and parallel processing for faster multi-user rendering.
© 2025 Kapsule Lab — Built for the Google Cloud Hackathon


Log in or sign up for Devpost to join the conversation.