Arunav Maitra

Posted on Sep 13

BrickVerse AI

#devchallenge #googleaichallenge #ai #gemini

This is a submission for the Google AI Studio Multimodal Challenge

What I Built

I built BrickVerse AI.

It's a magical portal where imagination meets digital creation.

This applet solves a simple, yet wonderful problem: How do you visualize any city in the world as a vibrant, intricate LEGO masterpiece?

BrickVerse AI creates a delightful experience by allowing anyone, regardless of artistic skill, to become a master LEGO builder.

You can start with just a simple city name.
Type "Paris"... and watch the Eiffel Tower rise, brick by brick.

Or, you can upload a personal photo.
A snapshot from your last vacation... and see it completely reimagined as a bustling LEGO world.

The goal is to spark creativity and bring a sense of childlike wonder to digital art, powered by the incredible capabilities of generative AI.

Demo

You can experience the magic live right here:
Link to Deployed Applet

Here’s a little sneak peek into the world of BrickVerse AI!

1. The Sleek & Simple Interface: Choose your creative path - text or image.

2. Generating from Text: We typed "Tokyo" and the AI started building...

3. The Final Masterpiece: A stunning, photorealistic LEGO Tokyo, complete with cherry blossoms and iconic towers.

How I Used Google AI Studio

Google AI Studio was my digital workshop for this project. It was the perfect environment to explore, prototype, and harness the power of Google's latest multimodal models.

I primarily leveraged two phenomenal models:

imagen-4.0-generate-001: For the text-to-image generation. I used AI Studio to fine-tune my prompts, experimenting with different keywords like "photorealistic", "cinematic lighting", and "bustling with LEGO pedestrians" to achieve that perfect, lively LEGO aesthetic. The ability to quickly iterate in the studio was invaluable.
gemini-2.5-flash-image-preview: This model is the heart of the image-to-image feature. My entire prompt engineering for transforming an existing photo into a LEGO world was done within AI Studio. I crafted instructions that guided the model to recreate, not just overlay, the source image, ensuring every building, tree, and car was reimagined in brick form.

AI Studio made the process of integrating these powerful AI capabilities seamless and intuitive.

Multimodal Features

The true essence of BrickVerse AI lies in its multimodal nature. It's not just about one input; it's about offering creative flexibility.

1. From Words to Worlds (Text-to-Image)

This feature allows users to conjure a world from pure imagination.

How it works: A user provides a text string (e.g., "New York City"). The application then embeds this into a more detailed prompt and sends it to the imagen-4.0-generate-001 model.
Why it enhances the experience: It's the ultimate creative sandbox. You don't need a reference; you just need an idea. It makes the creation process incredibly accessible and limitless. You can dream of a LEGO Venice during a flood or a futuristic LEGO Dubai, and the AI will build it for you.

2. From Pixels to Plastic (Image-to-Image)

This feature makes the creation process deeply personal.

How it works: A user uploads an image. The app sends this image data along with a text prompt (e.g., "Transform this entire image into a vibrant, detailed LEGO city scene") to the gemini-2.5-flash-image-preview model.
Why it enhances the experience: This is where the magic becomes personal. Users can upload photos of their own hometown, a favorite landmark, or a cherished vacation spot. The AI doesn't just add a filter; it understands the context of the image and rebuilds it. Seeing a personal memory transformed into a work of LEGO art creates a powerful and engaging emotional connection for the user.

By combining both text and image inputs, BrickVerse AI caters to different creative impulses, making it a truly versatile and captivating multimodal application.

DEV Community