Image generator

Lion in jungle, tiger by stream, romantic mood, captain observing, photorealistic, 8K details

Inspiration

We wanted to simplify the process of turning imagination into visuals. Many professionals struggle to find the right image that matches their creative vision. We were inspired to build a tool that could generate high-quality images from simple text prompts using generative AI.

What it does

Our platform takes a user’s text input and generates a contextually accurate, high-resolution image using a diffusion-based model (Amazon Titan via AWS Bedrock). It also allows users to view prompt history, download images, and manage image settings like size and format.

How we built it

Frontend: Built with React + Vite (TypeScript), clean and intuitive UI Backend: Python Flask API, handling requests and calling the Titan model via AWS SDK Model: Amazon Titan Image Generator (via Bedrock API) Deployment: Vercel (frontend), Render (backend) Security: Environment variables and secret key protection

Challenges we ran into

Secret scanning prevented GitHub pushes due to leaked AWS keys Vercel and Render networking issues (CORS, port bindings) Debugging build errors with vite.config.ts Learning to optimize AWS Bedrock model requests and JSON structures

Accomplishments that we're proud of

Successfully integrated a powerful GenAI model with our own UI/UX Full-stack deployment across two cloud platforms Clean, responsive UI with live rendering and history tracking Learned AWS Bedrock usage within days

What we learned

Working with foundation models like Amazon Titan Securely handling API keys and environment variables Bridging frontend-backend communication securely Real-time image generation with GenAI under production constraints

What's next for Image generator

Support for multiple models (like SDXL, DALL·E) Advanced prompt editing and negative prompts User login & personal image libraries Export to social media or design tools Mobile-first experience and offline capabilities