PRODUCT_BRIEF — BIM Cloud Pipeline

Objective

A proof-of-concept (POC) for a cloud BIM conversion pipeline that takes IFC (primary) and Revit-derived (optional, via Autodesk APS) BIM input, runs cloud processing jobs, and produces optimised GLB/GLTF 3D models plus structured BIM metadata JSON — making BIM geometry and semantics consumable on web, mobile, AR/VR, XR and digital-twin applications.

What this is not

A replacement for Revit, an engineering-analysis tool, a contractual BIM validation system, or a production multi-tenant SaaS. It is a reference implementation demonstrating the workflow.

Problem

BIM models are locked inside heavyweight, desktop-bound formats. Downstream consumers (web viewers, AR/VR headsets, digital twins, real-time AEC apps) need lightweight, runtime-friendly 3D assets plus queryable metadata. There is no single, simple pipeline that delivers both.

Solution

A single REST pipeline that: 1. Upload — accepts .ifc, .rvt, .gltf, .glb via dashboard or API. 2. Process — parses BIM geometry + semantics and builds derivatives. 3. Track — live job status, stage progress, and processing logs. 4. Deliver — model.glb (self-contained binary), model.gltf + .bin, and metadata.json (structured BIM data) with download endpoints. 5. Compare — side-by-side 3D plus an element/category diff of two models.

Target users & roles

• BIM / VDC teams — publish models for review and downstream use.
• Construction-technology developers — build web/mobile BIM apps on APIs.
• AR/VR / XR teams — get lightweight glTF assets for headsets & mobile.
• Digital-twin teams — feed geometry + metadata into twin platforms.

Primary journeys

1. Upload an IFC file → watch stages progress → download GLB + metadata JSON.
2. Run the bundled sample (one click) → inspect the 3D viewer + structured JSON.
3. Run two models and compare them (Architecture vs Structural).
4. Call the REST API to integrate the pipeline into a custom product.

Data (sample, open-source)

• samples/Building-Architecture.ifc and samples/Building-Structural.ifc — real IFC4 samples from the buildingSMART Sample-Test-Files repository (single-family house, architectural + structural discipline views) with walls, slabs, beams, roof, furniture, spaces, property sets, quantities, materials and classification.

Acceptance criteria (summary — see ACCEPTANCE_CRITERIA.md)

• Upload + process .ifc → real GLB + GLTF + metadata JSON.
• Job status, stage progress and logs are live.
• 3D model is viewable in-browser; metadata is structured and queryable.
• Model comparison works (Architecture vs Structural).
• Revit .rvt route is clearly handled: real Autodesk APS adapter (implemented, unit-tested with mocks, not live-validated). Live .rvt uploads are disabled by default (ALLOW_RVT_UPLOAD=0) to protect APS quota; the adapter is documented for local/self-hosted BYOC use.
• Responsive UI; no blocking console errors.

Capability status

Out of scope (production features intentionally omitted)

Multi-tenancy, authentication, billing, durable job queues, tenant storage, complex rate limiting, observability stacks and enterprise secret management.