Inspiration

Video content can boost conversion rates by up to 80%, but most ecommerce stores lack the resources to produce professional product videos. We wanted to democratize video commerce by turning existing product images into engaging videos automatically, without the need for cameras, studios, or production teams.

What it does

Our solution is a plug-and-play backend pipeline that transforms static product images into AI-generated videos using Higgsfield’s Image2Video API. Store owners simply integrate a plugin into their backend, and their image catalogs are instantly converted into engaging video experiences. The system works at scale, enabling thousands of products to generate videos automatically via webhook-based processing.

How we built it

Higgsfield’s Image2VideoAPI Next.js: Chosen for familiarity, powering the API server. Inngest: Handles asynchronous video generation tasks at scale. Supabase Storage: Stores both source images and generated videos. Plugin Integration: Works with Node.js, JavaScript, and even WordPress, making it easy for store owners to integrate. Demo: A Chrome extension to showcase real-time transformation of any e-commerce store’s product images into videos.

Challenges we ran into

Speed of video generation on Higgsfield’s Image2VideoAPI

Accomplishments that we're proud of

  1. Built a fully functional system that automatically converts static product images into engaging video content.
  2. Enabled small and large e-commerce stores to instantly offer immersive video shopping experiences.
  3. Developed a Chrome extension demo that effectively showcases the product’s potential.

What we learned

  1. Generating videos on-the-fly for every user is resource-intensive and wasteful.
  2. Without caching or shared storage, simultaneous usage by many users can slow down the system.
  3. Efficient backend design, including reusing previously generated videos, is critical for scalability and cost management.
  4. Thoughtful architecture (caching, storage, and asynchronous processing) is just as important as the AI generation itself.

What's next for Glow Up

  1. Implement a shared backend that stores previously generated videos, so new users can load existing content without regenerating it.
  2. Introduce intelligent caching and retrieval to reduce unnecessary AI processing and improve response times.

Built With

  • higgsfield
  • inngest
  • next
  • nextjs
  • supabase
Share this project:

Updates