Inspiration
Beauty and fashion are deeply personal, but the choices behind them often have environmental consequences that are easy to overlook. I wanted to explore whether a virtual try-on experience could do more than simply help someone preview a look. The idea behind EcoTry Bellini was simple: make beauty decisions feel more informed. Instead of only asking “How do I look?”, the project also asks “What kind of impact does this choice have?” I was also inspired by the challenge itself. Since the hackathon centered around AI-powered beauty and retail experiences, I wanted to build something that clearly used virtual try-on technology while adding a practical sustainability layer.
What it does
EcoTry Bellini is a lightweight virtual try-on experience built around occasion-based personalization. A user: Uploads a selfie Chooses an occasion such as Interview, Wedding, or Date Night Receives a virtual try-on preview powered by Perfect Corp’s technology Gets a short Sustainable Insight that suggests a lower-impact alternative with a similar aesthetic. The goal is not only to help users visualize a look, but also to encourage more thoughtful beauty choices.
How I built it
I built EcoTry Bellini as a lightweight web application designed for speed, clarity, and fast iteration under hackathon time constraints. As a solo participant, I handled the full stack of the application. Frontend I built the user interface with Streamlit and customized it with CSS using a Milanese gold and deep-charcoal aesthetic to create a luxury-inspired visual identity. AI Integration I integrated Perfect Corp’s Virtual Try-On (VTO) API so users could upload portraits and preview curated makeup looks based on their selected occasion. Backend and Security I developed a lightweight FastAPI backend to handle secure communication with external services. This kept API credentials out of the frontend and made the integration cleaner and safer. Sustainability Layer I built a custom logic layer that connects specific occasions—such as Interview, Wedding, and Date Night—with sustainability insights and lower-impact product alternatives that preserve a similar aesthetic. A major design goal was to keep the product simple enough to demonstrate clearly within hackathon constraints while still feeling polished, consumer-facing, and product-oriented.
Challenges I ran into
- The biggest hurdle was the "CORS Configuration Fight." Navigating Cross-Origin Resource Sharing and secure API communication while ensuring the frontend could correctly render cloud-processed images from yce.perfectcorp.com was a significant technical challenge.
- I also had to manage the high-fidelity data exchange between my Python backend and the Perfect Corp VTO engine. Ensuring the lipstick_id and color mapping coordinates were passed correctly while maintaining a responsive UI required deep-diving into network headers and request protocols to ensure every "200 OK" was earned.
- Data Synchronization Synchronizing the real-time AI image processing with the corresponding sustainability data required a rigorous approach to asynchronous requests. I had to ensure that the "Sustainable Insight" matched the visual output perfectly, maintaining the integrity of the information presented to the user. ## Accomplishments that I am proud of I am incredibly proud of taking a complex enterprise API and making it accessible through a "Sustainable Luxury" lens. Successfully implementing a secure handshake with the Perfect Corp AI Console as a solo developer is a major milestone. Additionally, I’m proud of the "Green Insight" logic—it’s one thing to make an app look good, but another to make it serve a higher purpose like environmental consciousness.
What I learned
This project taught me a great deal about API orchestration in a practical setting. I learned how to handle image-based API workflows, manage environment-based security for API keys, and translate a visual brand direction—Milanese gold and deep charcoal—into a functional Streamlit interface. I also gained a better understanding of deployment and network behavior during testing. Watching successful API calls and clean network responses in the browser inspection tools gave me a much clearer understanding of how frontend interactions, backend logic, and external services connect in a real application. More than anything, I learned that building a polished product under time pressure depends as much on clarity, testing, and iteration as it does on writing code.
What's next for EcoTry Bellini
The next step for EcoTry Bellini is turning the prototype into a more personalized beauty decision assistant. Future improvements would include: i. Broader occasion-aware recommendations beyond a small curated set ii. More personalized suggestions based on skin tone, style preferences, and past selections iii. Ingredient-level sustainability comparisons instead of general sustainability insights iv. Retailer integration so users can directly discover lower-impact alternatives A longer-term direction would be expanding beyond makeup into a wider beauty and fashion discovery experience, where virtual try-on helps users make more confident and more informed purchasing decisions. The long-term goal is simple: combine premium virtual beauty experiences with smarter sustainability awareness.
Log in or sign up for Devpost to join the conversation.