ReelMagic AI

Inspiration We were inspired by the growing accessibility of generative AI tools and the question: Can multiple AI models be orchestrated into a single, seamless creative pipeline? Creating short cinematic content typically requires scripting, voice recording, editing, and technical production skills. We wanted to explore whether this process could be simplified into a one-click experience powered entirely by AI. ReelMagic AI was our attempt to design and prototype that system during our first hackathon. What it does ReelMagic AI transforms a short text idea into an AI-generated video prototype. The user provides a simple concept (e.g., “A hidden city in the clouds”), and the system: Generates a short narration script Creates a cinematic video prompt Produces AI voice narration Calls a video generation model Orchestrates the flow through backend logic This MVP demonstrates the automated pipeline and model orchestration architecture. How we built it We implemented a full-stack architecture using: Next.js for frontend and backend API routes React for the user interface MiniMax M2.5 for script and prompt generation speech-2.6-hd for voice narration Hailuo 2.3 for video generation (asynchronous task polling) Vercel for development and testing The backend coordinates multiple AI calls in sequence. The most technically challenging part was handling asynchronous video generation, which required: Submitting a generation task Receiving a task ID Polling for completion Retrieving the generated result This architecture was designed under strict time constraints. Challenges we ran into This was our first hackathon experience. Music generation integration was not completed within the time limit. The video generation API returned results successfully, but the final video did not render properly in the frontend. Managing asynchronous polling correctly required careful debugging. API authentication and request formatting caused delays during development. We prioritized building a functioning orchestration pipeline over polishing all output features. Accomplishments that we're proud of Designing and implementing a multi-model AI orchestration system Successfully integrating LLM, speech, and video APIs Implementing asynchronous polling logic Building a working MVP within a limited time frame Structuring a scalable architecture for future expansion What we learned How to architect AI-driven pipelines How asynchronous generation systems operate The importance of structured backend orchestration The value of clear role distribution in a team How to adapt quickly under time pressure What's next for ReelMagic AI Complete music generation integration Fix frontend video rendering and output synchronization Implement automated audio/video merging Improve UI feedback and error handling Expand to multi-scene storytelling This project represents the foundation of a larger vision: a streamlined AI-powered cinematic content creation tool.

Built With

• vercel