Architexture AI

This is a submission for the Google AI Studio Multimodal Challenge

What I Built

Have you ever sketched a dream house on a napkin?
Or imagined a futuristic skyscraper that could reshape a city's skyline?

Bringing those ideas to life is often a monumental task.
It requires specialized software, technical skills, and a whole lot of time.

That's why I built Architexture AI.

It’s not just a tool; it's a creative partner for architects, designers, and dreamers. It closes the gap between your imagination and a stunning, visual reality.

The core experience is built around a simple, powerful, three-step loop:

1. Describe: You start with a simple text prompt. Pour your vision into words.
2. Generate: Instantly, Imagen 4 generates four distinct, high-quality architectural concepts based on your idea. No more blank canvas anxiety!
3. Refine: This is where the magic happens. Pick a design you like and start a conversation with it. Using the power of Gemini, you can ask for changes with simple text commands, iterating until it’s perfect.

Architexture AI makes architectural design fast, intuitive, and incredibly fun.

Demo:

live demo here, pls visit

Let's take a quick walk through the creative journey with Architexture AI.

1. The Spark of an Idea

Everything starts on our clean, focused welcome screen. You're presented with a simple text area, ready for your vision. We even provide a few examples to get your creative gears turning.

Let's try a prompt: "A modern eco-friendly villa with a green roof and an infinity pool overlooking a tropical beach."

2. AI-Powered Brainstorming

Once you hit "Generate," Imagen 4 gets to work. In moments, you're not just looking at one interpretation of your idea, but four unique, photorealistic concepts from different angles. This gives you a rich set of starting points to choose from.

3. The Creative Conversation

You've found a design that's close, but not quite it. Time to refine! Selecting an image takes you to the Editor. Here, you can simply tell the AI what you want to change.

Let's ask for an edit: "Change the time of day to a beautiful sunset."

4. Vision, Realized

With the power of Gemini's multimodal understanding, the AI doesn't just know what a sunset is; it understands how to apply that concept to your specific image. It considers the lighting, shadows, and reflections to deliver a breathtaking, context-aware result.

And just like that, an idea becomes a fully-realized vision, ready to be downloaded and shared.

How I Used Google AI Studio

Architexture AI is powered by a dynamic duo of Google's state-of-the-art models, orchestrated to create a seamless workflow.

1. Concept Generation with Imagen 4

For the initial generation phase, I chose Imagen 4. Its ability to produce high-quality, photorealistic, and creatively diverse images from a single text prompt is second to none.

I specifically prompt it to generate 4 different high-quality, photorealistic architectural visualizations... from multiple angles. This ensures the user receives a varied and inspiring set of initial concepts, which is crucial for the creative process. The API call is a straightforward use of ai.models.generateImages.

2. Iterative Editing with Gemini 2.5 Flash Image Preview

This is the heart of the multimodal experience. To power the editor, I used Gemini 2.5 Flash Image Preview (affectionately known as Nano Banana).

Its incredible strength lies in its ability to take both an image and a text prompt as input. The API call to ai.models.generateContent is structured with two parts:

• An inlineData part containing the base64-encoded original image.
• A text part containing the user's edit instruction (e.g., "add a swimming pool").

Gemini then generates a new image that incorporates the textual request while maintaining the context and style of the original image. This is what makes the iterative, conversational design process possible.

Multimodal Features

The true magic of Architexture AI lies in its multimodal capabilities, which fundamentally enhance the user experience.

• Conversational Design: The core feature is Image + Text → Image editing. This transforms the design process from a series of complex commands into a simple, natural conversation. Instead of fiddling with sliders and tools, you just ask for what you want. It feels like you're art-directing a creative assistant, not operating a piece of software.

• Context-Aware Creativity: By providing an image as context, the AI's response is grounded and relevant. When you ask to "add more windows," it understands the building's existing style, materials, and lighting, ensuring the edit feels seamless and natural. This is a massive leap beyond purely text-based image generation.

• Rapid, Risk-Free Iteration: Multimodality empowers users to experiment freely. Don't like the change? Just go back and try a different prompt. This frictionless workflow encourages creativity and allows for the rapid exploration of countless design variations without starting from scratch each time.

Ultimately, by combining visual input with textual instruction, Architexture AI creates an intuitive, powerful, and deeply engaging creative experience that makes the world of architectural design accessible to everyone.