This post is my submission for DEV Education Track: Build Apps with Google AI Studio.
TL;DR
I built a lightweight Avatar Profile Generator using Imagen inside Google AI Studio. Upload a face → get a clean, stylized avatar. The whole workflow runs in the browser using React + the Gemini API. Includes demo, source code, prompt, and lessons learned.
What I Built
I created an avatar profile generator that transforms a user‑uploaded face image into a clean, stylized social media avatar using Imagen. The goal was to show how easily multimodal inputs and generative outputs can be combined inside Google AI Studio.
The entire app runs in the browser using React and the @google/generative‑ai client, with a modular service layer that keeps the Gemini integration clean and extendable.
Prompt Used
This is the exact prompt powering the avatar transformation:
"Please create an app that takes an image of a face and generates an identical social media avatar using Imagen."
Keeping the prompt short and explicit helped the model preserve identity while applying a stylized look.
Demo
Live Project:
https://ai.studio/apps/f70cabcc-b531-4d55-b51a-9813f3569d0d?fullscreenApplet=true
Source Code:
https://github.com/lukeponga-dev/Face-Avatar-Generator
My Experience
Working through the Google AI Studio track gave me a much deeper appreciation for how quickly you can turn an idea into a working AI‑powered tool. A few things stood out:
1. Prompt design is a real skill
Small phrasing changes had a huge impact. Adding clarity, constraints, and examples dramatically improved consistency.
2. Imagen handles identity better than expected
It didn’t just apply a filter — it preserved facial structure while generating a stylized avatar. It genuinely understood the face.
3. The app‑building workflow is incredibly smooth
Building, testing, and deploying inside AI Studio felt lightweight and fast. Preview mode made iteration effortless.
4. Safety and guardrails are built in
Seeing safety checks handled automatically gave me a better understanding of responsible AI development.
5. Small projects can feel surprisingly complete
Connecting image input → model generation → output display made the app feel polished and usable.
Support the Project
If you found this useful, you can support the project by:
- ⭐ Starring the GitHub repo
- 🔁 Sharing the demo
- 💬 Leaving a comment or question below
I’d love to hear what you’d build with Imagen or Gemini.
Thanks to the DEV team and Google for putting this track together — it was a fun excuse to build something creative and ship it quickly.
Top comments (0)