Inspiration

We wanted to make party decoration easy, visual, and fun for everyone. Finding the right look for a room can be stressful, so we built an AI that does it for you. PartyDecor.AI helps people instantly see and plan beautiful setups without the guesswork.

What it does

Users upload a room photo, choose an occasion, style, and materials, and the AI decorates the space virtually. It generates a realistic edited image and a detailed step-by-step decoration plan. Everything fits your theme, lighting, and budget.

How we built it

We used Flask for the backend and HTML/CSS/JS for the frontend. Gemini 2.5 Flash Image handles realistic photo edits, while Gemini Text creates the decoration plan. The app integrates everything in one seamless, interactive UI with light/dark mode and gallery features.

Challenges we ran into

Getting Gemini to edit images realistically while preserving lighting and furniture was tough. Our biggest challenge, however, was integrating the Google API when we faced numerous 403 API Leaked errors, where using an alternative was recommended. When this was implemented, the same errors occurred. It took a while to steer back into a fully functional model with guidance. We also faced CORS and image encoding issues between frontend and backend. Lastly, conducting touch ups for the UI to be responsive and smooth under time pressure was thrilling but challenging.

Accomplishments that we're proud of

We built a fully working end-to-end AI decorator that feels intuitive and fun. The before/after visuals look realistic, and the text plans are practical. Our team collaborated efficiently, finishing a polished product within the hackathon deadline.

What we learned

We learned to integrate multimodal AI models and manage large image data efficiently. Prompt design turned out to be key for getting natural edits. We also gained experience in teamwork, UI polish, and real-world AI application building.

What's next for PartyDecor.AI

We plan to add AR visualization and voice-based editing (“move balloons to the left”). A built-in shopping list feature will help users buy materials easily. We also want to expand the gallery and let users share their designs online.

Share this project:

Updates