-
-
Home Page
-
Function1
-
Function2
Inspiration
Many social media creators struggle to create high-engagement covers that stand out. The "layered typography" (text sandwiched between subject and background) is a massive trend on platforms like Instagram and Xiaohongshu, but it usually requires professional Photoshop skills. We wanted to build a tool that makes this professional aesthetic accessible to everyone with one click.
What it does
RedVibe is a specialized AI visual editor that automates the "Sandwich Effect." It uses Gemini 3's multimodal intelligence to isolate subjects (people or objects) from the background and inserts customizable text into the middle layer. This creates a stunning 3D visual hierarchy that significantly boosts click-through rates for social media thumbnails.
How we built it
The application is built with a React frontend and a Node.js backend to ensure API security. We integrated Gemini 3 via the Google AI SDK to handle advanced image reasoning. The development workflow was accelerated by Trae, which helped us refactor the architecture from a local prototype to a secured, competition-ready global application in record time.
Challenges we ran into
One major challenge was ensuring the precision of subject extraction across diverse backgrounds. We had to fine-tune the multimodal prompts for Gemini 3 to accurately identify the "semantic layers" of an image. Additionally, managing the state between the text editor and the AI-generated layers required a robust state management solution.
Accomplishments that we're proud of
We successfully implemented a complex, layered rendering engine that works seamlessly with AI-generated masks. We are especially proud of the frictionless user experience—taking a professional design process that usually takes 20 minutes and shrinking it down to seconds.
What we learned
Building RedVibe taught us the immense power of Gemini 3’s Spatial Intelligence. We learned that AI isn't just for generating content, but for understanding the structural hierarchy of visual data, which opens up new possibilities for creative tools.
What's next for RedVibe
We plan to integrate RedVibe as a core visual engine for our AIbo (3.0 Avatar project). This will allow virtual identities to automatically generate branded social content. We also aim to add "Motion Sandwich" effects, allowing the text to animate behind subjects in short-form videos.
