Clonify

Logo
Created 3D Space
Account page
Space details
Landing Page

Inspiration

Urban spaces, public infrastructure, and residential buildings often lack accessible, low-cost spatial analysis tools. Conducting physical site surveys, assessing post-disaster structural integrity, or auditing public transit stops for ADA accessibility typically requires expensive CAD hardware or manual on-site teams.

We were inspired to build Clonify to democratize spatial AI—transforming standard 2D photos into high-precision, navigable 3D digital twins. By aligning our platform with the United Nations Sustainable Development Goals (SDGs 9, 10, 11, and 12), we set out to build a platform that streamlines disaster recovery assessments, eliminates carbon emissions from physical furniture staging, and automates accessibility compliance audits.

How We Built It

Clonify is engineered as a hybrid platform: lightweight enough for a fast live hackathon MVP, yet scalable for a production-ready architectural pipeline.

Architectural Core & Tech Stack

Frontend & Rendering: Built with Next.js (App Router), Tailwind CSS, and shadcn/ui. Interactive 3D web models rely on Google <model-viewer>, paired with Three.js / React Three Fiber for camera fly-through animations and custom mesh shaders.
Spatial & Indoor Mapping: Integrated the Mappedin SDK for multi-floor wayfinding and room layout generation, Archilogic API for 2D blueprint conversions, IndoorAtlas for high-precision indoor positioning, and the Overpass API (OpenStreetMap) for urban terrain contexts.
AI 3D Reconstruction: Powered by Luma AI (NeRFs / Gaussian Splatting) and Tripo3D / Meshy SDKs for asynchronous photo-to-3D mesh generation.
Backend & Data Pipeline: Driven by Prisma ORM with SQLite for zero-config local execution (upgradable to Supabase/PostgreSQL), AWS S3/Cloudflare R2 for asset storage, and Upstash Redis with BullMQ for asynchronous queue management.
Voice, Payments & Analytics: Web Speech API for spatial audio tours, Stripe API for Pro subscriptions, and PostHog for telemetry.

Challenges We Faced

Asynchronous Mesh Generation: Reconstructing NeRFs and Gaussian Splats from unstructured photo arrays is computationally intensive. Handling variable processing latencies required building robust background job processing with optimistic client-side polling.
Coordinate Alignment: Mapping 3D reconstructed meshes (.glb/.usdz) onto 2D/3D indoor floor plans generated by the Mappedin SDK required precise spatial coordinate normalization and scale calibration.
Real-Time Web Performance: Rendering high-density spatial meshes smoothly in the browser required optimizing geometry levels of detail (LOD) and shader execution.

What We Learned

Spatial AI Integration: Blending neural rendering pipelines (NeRFs/Splats) with traditional indoor mapping SDKs creates powerful workflows for real-world architectural applications.
Accessibility Automation: Translating physical compliance standards (like ADA slope and clearance rules) into mathematical algorithms on 3D meshes opens up scalable avenues for civic infrastructure planning.
Edge & Pipeline Architecture: Structuring a flexible database and queue design allows a project to run seamlessly in local hackathon environments while remaining instantly ready for enterprise deployment.