Bharat Vesh: Try On India’s Traditions with AI

This is a submission for the Google AI Studio Multimodal Challenge

What I Built

I built Bharat Vesh, an interactive and culturally rich AI applet that allows users to experience the beauty of Indian traditional attire. Users can upload a picture of themselves and, within moments, see their own image transformed into six different iconic Indian outfits, such as the Saree, Sherwani, and Kurta.

The applet provides a fun, accessible, and personal way for anyone to explore and celebrate India's diverse sartorial heritage. It's a virtual changing room that bridges the gap between curiosity and cultural appreciation, powered by Google's cutting-edge "nano banana" image editing model.

Demo

You can try out the Bharat Vesh applet here:
https://ai.studio/apps/drive/1VRzxNEb5GO1o5C0fPhnIz_T0avh2uA2d

Here are a few screenshots of the applet in action:

1. The Simple Upload Interface
Users are greeted with a clean and straightforward prompt to upload their photo.
[Image: Screenshot of the Bharat Vesh upload interface where a user can select a file.]

2. Original Photo and AI-Generated Results
Once an image is uploaded, the applet generates six variations, showcasing the user in different traditional dresses while preserving their likeness.
[Image: A grid showing the original user photo on one side and the six generated images in different traditional outfits on the other.]

3. A Closer Look at the Transformation
The model does an excellent job of integrating the clothing realistically onto the user's photo.
[Image: A side-by-side comparison of the original photo and one of the best-generated results, for example, the user in a detailed Saree or a regal Sherwani.]

How I Used Google AI Studio

Google AI Studio was the core of this project's development. I leveraged its powerful and intuitive interface to access and prototype with Google's latest multimodal models.

Specifically, I used AI Studio to:

1. Access the "nano banana" Model: The applet is built around this advanced image editing model. AI Studio provided a direct and seamless way to interact with its capabilities.
2. Engineer the Perfect Prompt: The magic of Bharat Vesh lies in the prompt. I iteratively designed and refined a prompt that instructs the model to take a user's input image, identify the person, and realistically "dress" them in a specified Indian outfit (e.g., "a vibrant Banarasi Saree," "an elegant Bandhgala Sherwani").
3. Rapid Prototyping: The "Run" feature in AI Studio allowed me to test different images and prompt variations in real-time, helping me fine-tune the instructions to achieve the most realistic lighting, fabric drape, and overall composition for the final applet.

Multimodal Features

The core of Bharat Vesh is its powerful multimodal functionality, which creates a highly personalized and engaging user experience.

The key multimodal feature is Image-to-Image Transformation Guided by Text. Here's how it works:

1. Image Input: The user provides a visual input—their own photograph.
2. Text Instruction: My code programmatically sends a text prompt to the model, like "Place this person in a traditional Lehenga Choli," along with the user's image.
3. Image Output: The model understands both the content of the image (the person's face, pose, and body shape) and the context of the text prompt. It then generates a new, edited image that merges these two modalities, resulting in a photorealistic depiction of the user in the new attire.

This enhances the user experience by moving beyond simple filters or text-to-image generation. Instead of creating a generic image, it transforms the user's own reality, making the result personal, surprising, and incredibly shareable. It allows users to see themselves as part of a rich cultural tapestry, which is a far more impactful experience.

From,
@akshatbindal
Currently Building with Nano-Banana