Nothing-to-wear

Inspiration

Fashion has always been more than clothing to me. The right outfit can shift energy, focus, and confidence for the entire day. But the daily friction — overbuying, mismatched wardrobes, uncertainty about fit — turns something inspiring into something inefficient.

The idea behind this project is simple: What if everyone had a personal styling and shopping assistant?

Not just a recommendation engine. Not just another chatbot.

A practical tool that helps you: - Buy smarter - Organize your wardrobe - Understand whether something fits your body before you purchase it - Experiment with outfits without waste

We started this at the hackathon because the foundational technology is finally mature enough. Generative AI can now move from novelty to real utility. My teammates share this belief — and the ambition to apply modern generative AI to everyday decisions that actually matter. We also see clear long-term marketing potential: fashion, sustainability, and personalization intersect in a powerful way.

What it does

We built a simple web application that enables virtual outfit experimentation. 1. The user uploads: - A full-body photo - A face photo 2. The system generates a personalized avatar. 3. The user can: - Add garments to a library - Use predefined pieces - Combine items into outfits - Try them on the avatar

It transforms “I don’t know if this will suit me” into a visual answer.

Beyond visualization, the system can also provide guidance on how to elevate an outfit.

How we built it

Given the hackathon constraints and limited after-work hours, we chose a pragmatic stack: - Frontend: React - Backend: Simple Node.js server

The focus was speed of iteration. Architecture decisions were guided by one principle: deliver a smooth user flow rather than over-engineer infrastructure.

Challenges we ran into

The most technically demanding part was avatar generation. Getting photorealistic, consistent results required:
- Careful prompt engineering
- Iterative experimentation
- Model comparison and quality validation
From a UX perspective, the challenge was different. We made a deliberate decision: We did not want to build another chat interface. Generative AI often defaults to conversational UI. But inside a product experience, users don’t always want to “talk.” They want to do. So we focused on embedding generative AI directly into structured flows. The AI works in the background. The interface stays clean and action-driven.

Accomplishments that we're proud of

- A smooth avatar creation flow
- Realistic, detailed avatars suitable for outfit try-on
- Fast iteration between multiple outfit combinations
- Style suggestions that feel additive rather than intrusive

What we learned

A key technical insight was understanding the difference between: - Gemini 3 Pro Image Preview - Gemini 2.5 Flash Image

We observed clear trade-offs in: - Output quality - Detail retention - Speed - Prompt sensitivity

Understanding these constraints helped us make informed product decisions rather than chasing idealized results.

What's next for Nothing-to-wear

This hackathon project is not a prototype for demo purposes only. It is first step of our MVP.

Next steps: - Refine avatar consistency and Improve garment realism - Integrate wardrobe organization logic - Build mobile application which will be available any moment to make your everyday solution for daily "nothing to wear" question

The long-term vision is to reduce unnecessary purchases, simplify styling decisions, and make sustainable fashion choices easier — not harder. Because “Nothing to wear” shouldn’t mean you need to buy more. It should mean you need better tools.