explaingit

deejoin/ifcdb

11Audience · generalComplexity · 3/5ActiveSetup · hard

TLDR

Product overview README for IFCDB, a cloud-native database that stores BIM data in IFC format and exposes it via SQL plus natural-language queries for construction projects.

Mindmap

mindmap
  root((IFCDB))
    Inputs
      IFC model files
      Natural language queries
      Property data
    Outputs
      3D visualizations
      Quantity takeoffs
      Clash reports
      Sub model IFC exports
    Use Cases
      Merge multi discipline BIM
      Run quantity takeoff
      Detect clashes
    Tech Stack
      NoSQL
      NewSQL
      SQL
      AI

Things people build with this

USE CASE 1

Merge sub-models from different construction disciplines into one queryable 3D view

USE CASE 2

Run a quantity takeoff across a BIM project to count rebar or material volumes in minutes

USE CASE 3

Detect collisions across a merged model and jump the viewer to each clash location

USE CASE 4

Export selected components from one or many models out as a fresh IFC file for downstream tools

Tech stack

NoSQLNewSQLSQLAI

Getting it running

Difficulty · hard Time to first run · 1day+

README is a product overview only with no install steps, license, or command examples.

In plain English

IFCDB is a database system built around the IFC format, the open standard used in the construction industry to describe buildings, their parts, and how those parts relate to each other. It is made by Shanghai DeeJoin Information Technology, and the README presents it as a way to keep an entire project's BIM data, which is short for Building Information Model, in one queryable place rather than scattered across separate model files. Under the hood the system uses a cloud native architecture that mixes a NoSQL store with a NewSQL store, then exposes a standard SQL interface on top so that attribute data, geometry, and the relations between building components can all be read with the same query language. The README also says IFCDB has an AI model wired in, so a user can type a request in natural language, for example asking for a count or a volume, and the system turns it into the right query. The README lists five features. Multi model visualization lets the user merge any number of sub models or models from different professions into one 3D view, with sectioning, measurement, transparency, and isolation controls. Properties and property set management stores attributes for every component and supports natural language queries such as range filters or counts by IFC type, with results that can be shown as pie charts. Sub model export lets the user pull components from one or many models and write them out as a fresh IFC file that other software can open or build on. Real time engineering quantity calculation runs the same kind of multi condition query but is aimed at quantity takeoff, the construction task of counting how much material a design needs. The README says this shortens manual takeoff work from hours to minutes and is especially suited to models with a lot of steel reinforcement bars. Collision detection runs across the full merged model using a spatial index, reports the location, the components involved, and the overlap volume, and can jump the 3D view to each clash. The README is short, lists no install steps, no license, and no command examples, so it reads more as a product overview than a developer guide.

Copy-paste prompts

Prompt 1
Summarize the five features of IFCDB and which BIM workflow each replaces
Prompt 2
Compare IFCDB's SQL-over-NoSQL-plus-NewSQL approach to a single Postgres BIM store
Prompt 3
Draft a sample natural language query I would type to count walls by IFC type
Prompt 4
Outline what install and license details the README is missing for a developer evaluation
Prompt 5
Sketch an architecture diagram of IFCDB based only on what the README describes
Open on GitHub → Explain another repo

Generated 2026-05-22 · Model: sonnet-4-6 · Verify against the repo before relying on details.