ebeela - the artificial intelligence, built for editing.

Inspiration

I wanted to move image editing from the "technical era" to the "intent era." Instead of spending hours learning complex masking tools and manual lighting adjustments, users should just tell the AI what they want. The inspiration was to build a fluid, professional workspace where the AI handles the "grunt work" of retouching and extraction.

What it does

ebeela is an AI-first editor that simplifies complex tasks. It allows you to:

  • Detach: Roughly highlight objects to remove them while the AI intelligently heals the background.
  • Place: Insert new objects into scenes with automatic, realistic lighting integration.
  • Edit Text: Modify text within existing images while maintaining the original font style and hierarchy.
  • Identity Management: Maintain consistent "Brands" and "People" identities via a simple @mention system.

How I built it

The app is built with React and a modular architecture for high performance, utilizing:

  • Gemini 3 (Reasoning): I use 'gemini-3-flash-preview' for the logic layer—identifying objects in messy highlights, labeling layers, and extracting text hierarchies.
  • Nano Banana Pro: I use 'gemini-3-pro-image-preview' for heavy lifting—generating 4K textures, generative inpainting, and lighting blending.
  • Local Storage: IndexedDB ensures projects, brands, and people are saved locally on the user's device for privacy and speed.

What I learned

  • Reasoning over Precision: With Gemini 3, the AI’s ability to understand intent is more valuable than the user's ability to create a perfect mask.
  • Visual Context is Key: I learned that providing the AI with specific visual context (like Brand screenshots) significantly boosts creative consistency.

What's next for ebeela

  • Generative Video: Integrating Veo 3.1 to allow users to animate canvas nodes.
  • Real-time Collaboration: Enabling teams to edit AI scenes together in a shared workspace.
  • Vector-to-Raster: Improving translation for higher-quality, scalable logo integration.

Built With

Share this project:

Updates