VidCraft: AWS & NVIDIA - use AI to give power to every voice

AI tool that transforms multiple clips and Creative Commons assets into polished video stories—automating scene mapping, narration, and editing using AWS + NVIDIA, ready to publish in minutes.

Comment 1

Inspiration

We were inspired by creators in underserved communities who have powerful stories but lack access to expensive production teams. We wanted to give them a way to create stunning, professional-quality videos using only their ideas and raw clips—no big studio required.

What it does

VidCraft uses NVIDIA AI and AWS infrastructure to transform raw user clips and Creative Commons assets into polished, narrated stories. It automatically maps scenes, finds the perfect stock footage or transitions, adds AI-generated narration, and outputs a final video ready to share.

How we built it

We built a React + Flask full-stack app that: • Uploads user clips to AWS S3 • Uses NVIDIA LLMs to generate a story structure and scene list • Retrieves Creative Commons assets to fill gaps between scenes • Generates voice narration using NVIDIA TTS • Uses ffmpeg to stitch everything into a seamless, polished video • Stores metadata in Snowflake for indexing and future remixing.

Challenges we ran into

• Stitching dynamic media with different aspect ratios and formats • Handling streaming AI generation and SSE updates smoothly • Balancing performance with creativity—making it fast without losing soul • Coordinating multiple cloud services (NVIDIA, AWS) reliably.

Accomplishments that we're proud of

• A fully automated storytelling pipeline powered by AI • Turning rough clips into something that looks like a produced short film • Building real-time progress updates to keep creators engaged • Creating a platform that empowers voices that might otherwise go unheard.

What we learned

• AI is powerful not just for replacing tasks but amplifying creative expression. • Orchestrating cloud AI services can be efficient when done modularly. • Even MVPs can deliver pro-grade storytelling when UX, media, and AI work together.

What's next for VidCraft: AWS & NVIDIA - use AI to give power to every voice

• Integrate custom machine learning and reinforcement learning to capture and preserve each creator’s unique style preferences. Over time, as creators use VidCraft, the system would learn their artistic signature—so that even in their absence, VidCraft could continue to create in the spirit and style of their original work, carrying forward their creative legacy. • Support for multi-language narration and subtitles • More creative styles and voice models • Fine-grained timeline editing in-browser • A community asset library powered by Snowflake vector search • Scaling the platform so every creator, anywhere in the world, can share their story with cinematic quality.

Share this project:

Updates