SpaceIdentity

• 

Inspiration

Everyone has stared at a room and thought: “This doesn’t feel like me.” Interior design is expensive, inaccessible, and often generic. We wanted to change that by creating an AI system that understands not only your space, but you — your personality, habits, aesthetic, lifestyle, and daily routines — and transforms it into a personalized 3D room that reflects your identity.

What It Does

SpaceIdentity generates a fully customized 3D “dream room” based on the user’s identity profile.

How it works:

Identity Quiz: Users answer questions about lifestyle, habits (night owl vs. early bird), tidiness level, color preferences, and aesthetic identity

Room Upload: Upload a photo or floorplan

AI Analysis: Gemini Vision detects furniture, layout, lighting, and dimensions

Personalized Design: Automatically generates an interactive 3D redesigned room tailored to the user’s identity

AI Reasoning: Explains exactly why each item and layout choice fits the user

From minimalist students to cozy maximalists, every design adapts to who you are — not just what looks nice.

How We Built It (High-Level Architecture)

Frontend:

Next.js 15, React, TailwindCSS

React Three Fiber + Drei for the 3D room viewer

Zustand for global state

Backend on Vultr Cloud Compute:

Ubuntu 22.04 VM running Docker + Nginx reverse proxy

Identity quiz engine

Room analysis API

Gemini AI orchestration

3D layout generation service

Secure server environment (UFW firewall, SSL, non-root containers)

Vultr Object Storage:

Stores GLB 3D models (bed, desk, dresser, nightstand, plants)

Stores user-uploaded room images

Serves optimized assets via S3-compatible API

Custom bucket: dreamroom-assets/models/, dreamroom-assets/rooms/

AI:

Google Gemini 2.0 Flash (Vision + Text)

Used for room parsing, style detection, identity mapping, and layout reasoning

Deployment: One-command script (./vultr-deploy.sh) performs:

System provisioning

Docker image builds

SSL setup

Firewall config

Automatic service restart + orchestration

Why Vultr

We didn’t just “host” our app on Vultr. Vultr is the engine behind the entire system.

Vultr Cloud Compute powers:

Real-time AI orchestration

Layout generation

Quiz logic

3D model pipeline

All API routing (via Nginx)

Vultr Object Storage powers:

High-speed CDN-like delivery of GLB furniture models

Storage for user room images

S3-compatible access from our Next.js backend

Cross-origin 3D asset serving for the Three.js viewer

The whole system depends on Vultr’s reliability, speed, and flexible S3 tooling to create an identity-driven design experience that feels instant and immersive.

Challenges We Ran Into

Normalizing GLB furniture scales from different sources

Getting consistent room/lifestyle reasoning from Gemini Vision

Serving GLB assets cross-origin (CORS setup on Vultr Object Storage)

Optimizing 3D scenes for smooth rendering

Reducing Docker image size from 2GB to under 500MB

Accomplishments We’re Proud Of

Truly identity-first design generation

Full-stack deployment on Vultr (compute + storage)

Real-time 3D visualization that works across devices

Transparent AI reasoning that explains design decisions

Sub-3-second room analysis and regeneration

What We Learned

Vultr’s cloud ecosystem is powerful for rapid AI deployment

S3-compatible storage makes asset delivery extremely flexible

Docker + Vultr = simple, smooth, reproducible deployments

Personalization is much stronger when tied to identity, not only style

What’s Next

Shopping integration for furniture recommendations

AR room preview

Multi-room and whole-home design

Designer collaboration mode

Vultr-based analytics dashboards

Built With

• aws-sdk
• docker
• docker-compose
• drei
• google-gemini-2.0-flash-(vision-+-text)
• next.js-15
• nginx
• react
• react-three-fiber
• tailwindcss
• three.js
• typescript
• ubuntu-22.04
• ufw-firewall
• vultr-cloud-compute
• vultr-object-storage
• zustand