Morphica - AI Profile Picture Generator

Inspiration

Creating high-quality, personalized profile pictures often requires design skills or expensive tools. We set out to build an AI-powered application that makes this process effortless, allowing users to generate stylized avatars using cutting-edge generative models. Morphica was developed to simplify identity expression through technology.

Morphica – AI Profile Picture Generator enables users to upload an image and apply custom styles using generative AI. It also supports prompt-based image generation without requiring an upload.

Architecture and Tools

Frontend: Streamlit for a responsive and interactive UI. Backend: Deployed on AWS EC2 with a virtual Python environment. Generative AI: Amazon Bedrock’s Titan and Stable Diffusion XL models for image-to-image and text-to-image generation. Deployment: Managed using SSH, SCP, and Python virtual environments.

Key Features

Upload an image and apply a custom style prompt. Generate profile pictures in real time. Supports both image-to-image and text-to-image modes. Fully deployed and accessible via public EC2 URL.

What We Learned

Practical use of Amazon Bedrock for image generation tasks. Deployment and security management on AWS EC2. Building user-focused applications with Streamlit. Efficient team collaboration with Git and CLI-based workflows.

Challenges

Understanding and formatting Bedrock API requests for image transformations. Managing secure deployment and key-based authentication. Optimizing performance and stability for real-time generation.

Future Enhancements

User authentication and saved image history. Improved prompt suggestions via LLM integration. Upscaling and watermark-free download options. Transition to serverless architecture for scalability.

Built With

• amazon-bedrock
• amazon-ec2
• boto3
• git
• github
• python
• shell
• streamlit