Data formats for exchanging and manipulating 3D city models: Keep it simple

Keynote given at the DTCC Day 2021

https://dtcc.chalmers.se/

Transcript

1. Hugo Ledoux
   
   
   Data formats for exchanging and
   manipulating 3D city models:
   
   
   keep it simple
   Digital Twin Cities Centre Day
   
   
   2021-12-07
   Delft University of Technology
   

   View Slide

2. Topic of my talk: which format should we use?
   2
   .obj ?
   
   
   .gml ?
   
   
   .json ?
   
   
   .shp ?
   
   
   .stl ?
   3D city models of
   
   
   large area
   with textures with semantics
   with attributes
   

   View Slide

3. We want to process and exchange the information
   3
   • ideally directly on the web
   
   
   • with several different software
   
   
   • AutoCAD, GIS, Blender, Rhino, etc
   solar potential
   shadow
   analysis
   bomb detonation
   wind turbulence
   spatial planning
   

   View Slide

4. What are our options?
   4
   OBJ/STL/PLY
   
   
   COLLADA
   glTF + b3dm
   CityGML
   
   
   LandInfra + InfraGML
   CityJSON
   { ✅ very simple
   
   
   ✅ used for 30y+ in computer graphics + engineering
   
   
   ✅ zillions of software
   
   
   ❌ no semantics, no complex geometry, no attributes
   ✅ for visualisation on the web
   
   
   ✅ great software support
   
   
   🤨 simple-ish (binary)
   
   
   ❌ no semantics, no complex geometry
   ✅ OGC standards since 2008/2016
   
   
   ✅ do everything!
   
   
   ✅ *seem* perfect
   
   
   🤨 is it the case in practice?
   🚀 what I propose, more details in a few slides
   

   View Slide

5. 5
   International standard (from OGC) for representing
   and storing 3D city models
   

   View Slide

6. CityGML files are very complex
   6
   • files are deeply nested
   
   
   • aims at solving many problems
   
   
   • many “points of entry”
   
   
   • many di
   ff
   ways to do one thing
   

   View Slide

7. Consequence #1: almost no File/Open CityGML
   7
   

   View Slide

8. “GML madness”: how many ways to store a simple square?
   8
   25 different
   ways
   

   View Slide

9. Consequence #2: people will not download your files
   9
   -Raphaël Bovier, from his presentation at the 3D city modelling Workshop, Nordic
   Cooperation in the
   fi
   elds of Spatial Data and Land Administration (2021-09-29)
   

   View Slide

10. Consequence #3: no web support
    10
    No known JavaScript
    parser for CityGML files!?
    

    View Slide

11. “But my CityGML dataset is on the web!?”
    11
    

    View Slide

12. Consequence #4: developers/users will not use your format
    12
    GML 3.2.2 specs
    
    
    (427 pages)
    GeoJSON specs
    
    
    (28 pages )
    

    View Slide

13. Consequence #5: scientists in other communities are not involved
    13
    • CFD/simulation community == OBJ/STL
    
    
    • Computer Graphics == OBJ/STL
    
    
    • Other communities have scientists involved in the
    ecosystem
    

    View Slide

14. Consequence #6: alienate potential users (eg students)
    14
    When learning about semantic
    3D city models
    Trying to read a CityGML file
    
    
    with Python
    

    View Slide

15. Shouldn’t we apply the Rule of Least Power?
    15
    

    View Slide

16. View Slide

17. CityJSON v1.1
    17
    • community standard
    
    
    • compliant with CityGML v3.0
    
    
    • subset of CityGML (~90% of features)
    
    
    • software for full conversion CityGML <-> CityJSON
    
    
    • several software support CityJSON
    new!
    new!
    

    View Slide

18. Same information as CityGML, but in JSON format
    18
    {
    
    
    "type": “CityJSON",
    
    
    "version": “1.1”,
    
    
    "metadata": {
    
    
    "referenceSystem": "https://www.opengis.net/def/crs/EPSG/0/7415",
    
    
    },
    
    
    "CityObjects": {
    
    
    "id-1": {
    
    
    "type": "Building",
    
    
    "attributes": {
    
    
    "measuredHeight": 22.3,
    
    
    "roofType": "gable",
    
    
    "owner": “Elvis Presley"
    
    
    },
    
    
    "geometry": [
    
    
    {
    
    
    "type": "MultiSurface",
    
    
    "boundaries": [
    
    
    [[0, 3, 2, 1]], [[4, 5, 6, 7]], [[0, 1, 5, 4]]
    
    
    ]
    
    
    }
    
    
    ]
    
    
    }
    
    
    },
    
    
    "vertices": [
    
    
    [23.1, 2321.2, 11.0],
    
    
    [111.1, 321.1, 12.0],
    
    
    ...
    
    
    ],
    
    
    "appearance": {
    
    
    "materials": [],
    
    
    "textures":[],
    
    
    "vertices-texture": []
    
    
    }
    
    
    }
    
    
    human-readable file
    
    
    computers prefer this over XML
    
    
    web prefers this over XML
    
    
    ~6X compacter than CityGML
    

    View Slide

19. Python parsing is very easy
    19
    

    View Slide

20. Compression of files: Zürich LoD2 buildings
    20
    CityGML = 3.0GB
    
    
    (but 1GB of spaces/CRs/tabs!)
    CityJSON = 292MB
    

    View Slide

21. CityJSON can be easily parsed with JavaScript
    21
    Developed by MSc
    students
    

    View Slide

22. CityJSON is “ready” for machine learning
    22
    new!
    

    View Slide

23. 3D BAG: all 10M+ buildings in the Netherlands
    23
    https://3dbag.nl
    

    View Slide

24. LoD 2.2
    

    reconstruction
    3D BAG is based on open datasets
    24
    Point cloud
    (AHN3)
    Footprint (BAG)
    +
    X 10,000,000
    
    
    100% automatic
    

    View Slide

25. View Slide

26. Downloads in different formats
    26
    Stats October/November
    
    
    (~158k downloads; ~2.6k/day)
    
    
    CityJSON 72%
    
    
    OBJ 16%
    
    
    GPKG 11%
    
    
    PostgreSQL <1%
    

    View Slide

27. thank you.
    h.ledoux@tudelft.nl
    
    
    3d.bk.tudelft.nl/hledoux
    Hugo Ledoux
    https://cityjson.org
    https://3dbag.nl
    

    View Slide