How to Easily Read and Write CityGML Data (Without Coding)

How to Easily Read and Write
CityGML Data (Without Coding)

Meet the Presenters
Don Murray
Co-Founder
Dean Hintz
Product Manager, XML
Team & Open Standards
To chat with about anything data, anytime info@safe.com

Agenda
What weʼre covering today.
● Setting the Scene: CityGML context
● CityGML as Digital Twin
● CityJSON and CityGML
● Typical CityGML Workflows
● FME and CityGML, Indoor, BIM, and 3D
● Demos
○ Inspection, GeoBIM, Indoor, and more
● Trends and Conclusions
● Next Steps

Setting the Scene
World of 3D, Data, BIM, GEOBIM etc.
● Where does CityGML play?
● What are itʼs strengths,
● What are itʼs limitations?

CityGML Context: Strengths and Limitations
Strengths
● Data exchange format for 3D GIS, BIM-lite data
● Open Standard with solid cross industry adoption
● Object oriented model with diﬀerent LODs - more than just a pretty picture
● Semantic Models satisfy thematic queries, analysis, spatial data mining
● Facilities management applications, City scale, great source for digital twin
Limitations
● Doesnʼt scale well - XML, no spatial index, not optimized for storage
● Complex data model (need data model oriented tools like FME)
● Not a client oriented format. Diﬀicult to stream. Need conversion to web and mobile
oriented formats like PDF, 3js, GeoJSON
● Not heavy enough for true BIM - community and city scale is the main focus

CityGML Structure: Module Organization
© KIT – Karlsruher Institute of Technology

CityGML Structure: Semantics and Geometry
(C) by T.H. Kolbe
© T.H. Kolbe
© Filip Biljecki, Hugo Ledoux, Jantien Stoter

Setting the Scene PT II
Common cases where CityGML adds value.
● Digital Twin
● 3D GIS for City Systems and Urban planning

CityGML as Digital Twin
“A digital twin is a representation that serves as the real-time digital counterpart of a
physical object or process” - Wikipedia
● Digital twin is more than just 3D, itʼs about modelling real world objects
● CityGML provides the rich object model needed so digital objects have a
representation of the real world objects it is modelling.
● CityGML & CityJSON: Bridge the gap between GIS and BIM - Tie the detail of BIM to
the locational awareness of GIS.
● Semantic richness of feature types & properties allows CityGML to model down to
wall materials, building installation components and city furniture elements (lamp
posts and park benches).

CityJSON and CityGML
CityJSON advantages
● Smaller - about ⅙ the size
● No application schema (good and bad)
● Simpler geometry implementation
● Better for thinner clients
CityGML advantages
● Schema supports validation of both geom and schema
● Easily extensible with ADEʼs via GML app schema
● More comprehensive implementation of the CityGML standard
● Richer object model, ability to define xlinks

Typical CityGML Workﬂows
● Data Inspection and Evaluation
● CityGML model generation
● Creating LODs
● 3D data generation from 2D
● 2.5D draping
● 3D extrusion
● Database loading
● BIM to GIS
● CityGML to CityJSON
● Export to client formats (PDF/KML)

With FME you can easily exchange data
for visualizations, spatial analysis and
projects in CityGML.

FME®
Enterprise Integration Platform
FME Desktop
Build & Run Data Workflows
FME Server
Automate Data Workflows
Can be hosted on FME Cloud
FME Data Express
Powering your dataflows on premises, in the cloud, and on mobile
The only Enterprise Integration Solution with comprehensive support for Spatial Data
FME AR

Spatial Data is Everywhere...
... and only supported by FME.

Our Strength
We deliver the power of spatial data to the
worldʼs decision makers.

Connect
Data between
450+ systems
Extend
Capabilities with custom
connections, transformers,
R & Python
Run
Workflows locally or publish
to FME Server
FME Desktop
Quickly author repeatable, self-documenting data integration
workflows using an intuitive visual interface.
No coding required.

FME & CityGML - Highlights and
Recent Enhancements

FME Enhancements: CityGML
● CityJSON
● CityGML v3* (via GML R/W)
● Geometry by reference (xlinked)
● New LODs to Read parameter
● Coordinate precision parameter on Writer
● Improved geometry support
● Performance optimizations
● GML core improvements
● OGC Certification (GML but no test suite for CityGML yet)
*In progress

FME enhancements: Indoor, BIM and 3D
● IFC reading / writing improvements
● Native Revit Reader enhancements
● Destination format improvements: I3S, Cesium, other 3D, Indoor mapping etc
● New GeoBIM related formats: CityJSON, CityGML3, BCF, InfraGML etc
● New 3D formats: AR, Unreal, USD Pixar, Rhino3D etc
● Improved geometry handling and performance (GeometryValidator,
CenterLineReplacer, Clipper (surface clipping))
19

CityGML Demos
● Data Inspection: Montreal, NYC, Hi rise (geobim bm), CityJSON, reference geometry
● CityGML version 3
● Publishing to client application formats (CityGML to 3DPDF)
● GeoBIM benchmark: BIM to CityGML, CityGML to BIM ( examples)
● Indoor GML Pilot: CityGML as exchange format:
○ IFC > CityGML > IndoorGML
○ 2.5D to 3D (IMDF to CityGML)
● CityGML & FME Server - 3D Geoportals: CityGML download

CityGML Demos - cont’d
● CityGML to CityJSON
● Database loading: CityGML to GIS database (geodb)
● Automations and web 3D
○ Validation of CityGML
○ CityGML to Cesium / I3S
Some of these will be available as part of the download package after the webinar as we likely
wonʼt have time for them all today

Visualizing and Inspecting CityGML
and CityJSON

CityGML Inspection & Visualization: Montreal

CityJSON Inspection & Visualization: Rotterdam

CityJSON Inspection & Visualization: Railways

Geometry By Reference: xlink:href
Building geometry
links to component
walls, roof, floor

CityGML 3
● More flexible LOD & geometry model, addition of point clouds
● Time and versioning
● BIM lite - support for Spaces - better interop with IFC
● Indoor friendly - addition of building units and storeys (level) - better interop with
INSPIRE and IndoorGML
● Improvements to transportation objects for traﬀic, and driving simulations &
autonomous systems, road and railway facility management
● Separate conceptual model from encoding (GML, JSON and DB likely)
○ https://github.com/opengeospatial/CityGML-3.0Encodings
○ https://link.springer.com/article/10.1007/s41064-020-00095-z

CityGML3 with GML Reader: Transportation - OpenDRIVE

Publishing CityGML to Client
Formats: 3DPDF

CityGML to PDF: Montreal
Geometry Transform: deaggregate, triangulate, convert to mesh, group
Schema Transform: Expose, create attributes and set geometry traits

GeoBIM Benchmark
Purpose: Evaluate maturity of GeoBIM standards
and tool implementations
IFC and CityGML Testbed
● Task 1: IFC
● Task 2: IFC Georeferencing
● Task 3: CityGML
● Task 4: Conversion
Lead research coordinator: Francesca Noardo, TU Delft
Sponsored by ISPRS, EuroSDR
https://3d.bk.tudelft.nl/projects/geobim-benchmark/

● Open standard for Building Information Models
● Both STEP and XML
● Added support for the upcoming IFC 4x3
● Working with buildingSMART on test beds
CityGML
● Open standard for city information models
● FME has unparalleled support for CityGML ( v 1,2,3)
● Full support for ADEʼs (Energy, Noise, IMGeo)
● Recently added CityJSON support
● Active with CityGML SWG at OGC
IFC

IFC & CityGML: Modelling Challenge
*Source - slide from: Thomas H. Kolbe - joint work with Claus Nagel & Alexandra Stadler

Method: Read / write GML using CityGML application schemas & ADEʼs
Results:
● Able to read all the datasets evaluated
● Read and wrote all element types
● Good georeferencing support
Challenges:
● Rotterdam: Data Quality issues - surface errors within Building,BuildingPart.
Textures incorrect - errors: Non-Planar surfaces, and Surface not closed
(GeometryValidator)
● Possibility for user error - id, parent ids, semantics; size of schema
Task 3: CityGML

Amsterdam LOD1 CityGML Dataset

Geometry Quality - Rotterdam LOD1, LOD2 CityGML

Rotterdam LOD1, LOD2 CityGML Dataset

Rotterdam LOD1, LOD2 CityGML - Geolocation

IfcWall to CityGML WallSurface
ConvertGeometry custom transformer

Task 4: IFC to CityGML
Input IFC
Output CityGML

GeoBIM Benchmark - Lessons Learned
IFC
● Overall good reading support, diﬀerent data views
● Georeferencing may require workspace configuration
● Complex workflow typical for writing (geometry conversion, resolve relationships, set properties)
CityGML
● Good reading, writing and georeferencing support
● Complex workflow typical for writing (geometry conversion, resolve relationships, set properties)
Conversion
● BIM to GIS successful, - georeferencing and semantics requires configuration
● GIS to BIM, BIM to BIM (updates) challenging due to assumptions going from simpler to more
complex model
● Model based approach - complex transformation model, but provides more flexibility, control and
more easily supports automation

Supporting Indoor Mapping with
CityGML: IndoorGML Pilot

Indoor Mapping
IMDF ( Developed by Apple)
● OGC Open Standard (JSON)
● Strict Business Rules
● Designed for map display
ArcGIS Indoors
● Esriʼs Indoor Mapping System
● Designed for data storage/editing
IndoorGML
● OGC Open Standard (GML)
● Designed for data transfer

OGC Indoor GML Pilot: Design
Goal: Convert building interior scans into standardized building datasets able to support
indoor mapping
Design: Read LIDAR scans and convert to OGC IndoorGML for use by navigation services and
clients
1. Read LIDAR point clouds
2. Interpret features relevant to navigation: walls, doors, rooms
3. Generate CityGML with spaces and features
4. Enrich with public safety information (Public Safety ADE)
5. Convert CityGML to IndoorGML (PS extension)
6. Add navigation network to IndoorGML state and transition layers
7. Post results to CSW; Create a navigation service on top
8. Create clients to interact with navigation service

OGC Indoor GML Pilot: Implementation
● Public Safety CityGML ADE and Indoor GML PS extension had to be developed and
tested.
○ In parallel with PS extension development, early testing conducted with
CityGML and IndoorGML core
● Multiple tools and platforms to filter and clean LIDAR scans, convert to 3D geometries,
and simplify:
○ TICA with GML (direct edit of GML)
○ Revit with IFC (extra conversion step)
● Once data was in a standard geospatial format (CityGML or IFC), conversion to
IndoorGML was automated, after model based conversions were configured and tested

IMDF to CityGML: Victoria Airport Source IMDF

IMDF to CityGML: Victoria Airport Workspace

IMDF to CityGML: Victoria Airport CityGML Out

Victoria Airport Public Safety ADE

Indoor GML Pilot:
CityGML for data exchange
IFC to CityGML to IndoorGML

Indoor Pilot: Source - IfcSpace

Indoor Pilot: IFC to CityGML Workspace

IFC to CityGML: Geometry Transform & Clean
Filter complex objects to get just vertical walls
● Compute surface normal on all surfaces
● -0.1 < surfaceZ < 0.1 produced vertical wall faces
● surfaceZ > 0.5 produced the top of floors
● surfaceZ < - 0.5 produced the bottom of the roof or ceiling

GeometryPartExtractor GQuery
ConvertGeometry custom transformer
IFC to CityGML: Geometry Transform
Results were suﬀicient to generate CityGML from IFC

Indoor Pilot: CityGML to IndoorGML
FME Workspace

Geometry Transformations
Dissolve, Generalize, Snap, Aggregate

Schema Transformations
Attribute Mapping Value Mapping Conditional Mapping

Indoor Pilot Navigation: Network
IndoorGML States and Transitions

Indoor Pilot Navigation:
ShortestPathFinder
From – To vector and associated shortest
path

Navigation Modeller: Results
● CityGML Public Safety ADE testing / validation
● Automated Indoor GML generation from CityGML
● Demonstrate the use of the standard for OGC community
● Public safety planners to assist emergency response
● Development of FME conversion models:
○ IFC to CityGML transform
○ CityGML ADE to IndoorGML PSX transform
○ Navigation network generation
○ Validator

IndoorGML Pilot Lessons Learned
● It is feasible to convert from LIDAR to IFC, CityGML and IndoorGML
● Extraction of features / spaces from PC a challenge for automation, though some
methods hold promise
● Model based transformation tools support rapid testing and prototyping of new
schemas. Once workflow was developed, easy to apply to new building datasets.
● IndoorGML data model complexity poses implementation challenges. Still, schema
enables validation automation
● Accurate doors critical to correct navigation network. In some cases misclassified
doors had to be reclassified.
● Space concept is vital for integration across standards.

CityGML to CityJSON
https://www.cityjson.org/specs/1.1.0/
https://www.cityjson.org/conformance/v20/
Simplified property and
geometry model
Walls, Roofs etc are
children of Building and
written as BuildingPart
gml_id > fid
gml_parent_id >
cityjson_parents{0}
Building: cityjson_children{}

CityGML to Geodatabase
Flatten property and geometry
model
GeometryCoercer to
fme_composite_surface
(geodb_multipatch)
BulkAttributeRenamer to
convert . to _
GeometryPropertyExtractor -
prefix traits = yes
AttributeExposer + feature cache

CityGML Stored in Geodatabase
Flattened property and
geometry model -
Geodb_multipatch
Spatially indexed
CityGML and geometry
properties stored as fields
in database record
Need to rebuild object
hierarchy to write back to
CityGML

Partners
● ESRI: CityGML to I3S: Streaming fast 3D city data in ArcGIS
● Autodesk: FME CityGML Importer: Autodesk
● con terra:
○ Using CityGML in ArcGIS
○ https://www.conterra.de/bim (German)
● virtualcitySYSTEMS:
○ Spotlight
○ VC Warehouse

Customer Use Cases
City of Karlsruhe
● Generation of CityGML data using FME
New York City
● AppGeo's NYC project
● FME UC 2016: Stranger in a Strange Land;Exploring 3D and CityGML
Berlin 3D City Download Portal

con terra: Building Information Service
● ArcGIS Scene Service based 3D CityModel (originally
based on CityGML)
● For all Germany (> 50 million buildings)
● Integrates CityGML data into the ArcGIS Platform
● Quality assurance and improvements
● enhancing the model through e.g. volume and height
calculations
● Typical use cases:
● Building valuation for insurance
● urban planning
● telecommunications (antenna planning)

Berlin 3D Download Portal: FME Server
Download options:
Extents
mesh or LOD model
Model: (textures, tiles, generic
attributes)
Format:
CityGML, PDF, KMZ, SHP, 3DS, OBJ,
COLLADA, DWG, GeoDB, Geopackage,
IFC, FMEAR, FBX, GLTF, STL

Berlin 3D Download Portal: FME Server

The Goal
NYCʼs IT (DoITT)
needed a 3D
building massing
model for the Cityʼs
1,000,000 buildings
The Obstacles
Massive volumes of
data and huge data
quality challenges
The Solution
AppGeo used FME to
model all buildings
in 3D CityGML.
AppGeo then
automated analysis
of the CityGML
making QA reports.
The Result
Access to 3D
building models,
planimetric data
along with viewing
tools is now
provided by the
City.

The Goal
Property Register
Department
needed an eﬀicient
and automated
way of processing
the 3D data it had
collected
The Obstacles
Massive volumes of
data: urban area of
173 km², with 27
districts and a total
population of
approximately
300,000.
The Solution
Building data is
enriched and then
converted into
CityGML format & run
repeatedly each time
data isrequired.
The Result
Now able to
transfer 3D data
quickly and
eﬀiciently into
CityGML using a
standardised
workflow.

Tips and Tricks
● Upgrade to FME 2022.0 build 22270+ (2 X performance improvement for textured data
- Montreal reads in 1:30 now, before was > 3:00)
● Simplify geometry for downstream applications when complexity not needed. Use
LOD on reader.
● Use GeometryPartExtractor + GQuery to remove unnecessary geometry parts
● Reproject to a projected CRS like mercator (not degree based)
● Use CityGML relationships to join and group objects. For example gml_parent_id
● Work on 1 feature type at a time to speed development and minimize diagnostics
● Look at the structure of existing well formed citygml dataset when learning to write
● GeometryPropertyExtractor prefix traits=yes, AttributeExposer import from cache
● Configure FME for GPU

AttributeExposer > Import from Feature Cache
Choose: Import from Feature Cache > Selected desired properties > Auto-added to schema

NVIDIA Control Panel for Data Inspector GPU
Choose: High Performance Processor

CityGML Writing Tips
● Prepare your source data: geometry and source attributes
● Add CityGML specific attributes and geometry properties
○ Format Attribute citygml_lod_name
○ lodxSolid[1..4], lodxMultiSurface[2..4], lodxGeometry[0..5]
● Build relationship between features
○ Example: 'Building' address 'Address'
○ Use gml_id and gml_parent_id for relation
○ Format Attribute citygml_feature_role
● Model CityGML entities by FME feature types
○ Import CityGML entities from writer or from an existing CityGML dataset
○ Make sure you use exact case for all property names boundedBy ≠ Boundedby

CityGML Demos & Resources - from Safe
● CityGML on Safe.com
● Writing CityGML tutorial
● FME & 3D Article
● 3D Tutorial
● BIM tutorial (IFC toi CityGML)
● BIM to GIS Advanced
● Indoor GML Pilot

CityGML Demos & Resources - more
● con terra & Virtual City Systems: FMEHub CityGML Solution Templates
● Awesome CityGML https://github.com/OloOcki/awesome-citygml
● Simplified CityGML-Base 3d Space Model for Indoor Applications publication
● CityGML to I3S: Streaming fast 3D city data in ArcGIS
● From CityGML 3D City Model to 3D Web Application (CityGML to Cesium with FME)
● Safe CityGML Jan 2022 blog
● https://towardsdatascience.com/open-source-3d-semantical-building-models-in-2020-
f47c91f6cd97

Trends: CityGIS, BIM, 3D
● Convergence in 3D domain: 3D web tiles (I3S, Cesium), gaming, AR, simulation –
improved simplification processes
● More eﬀorts are being made to improve integration between BIM & GIS (GeoBIM: IFC
& CityGML), extract from BIM to lighter profiles or GIS, CityGML3, InfraGML
● Trend towards enterprise-wide integration, with workflow automation and updates
flowing across systems and between departments
● Demand for 3D city data is exploding. Challenge will be to keep up with this –
solutions need to be easy to deploy and flexible – cloud and mobile support.
● Updates are constant. Volumes are massive. Scalability and automation are key.

Conclusions
Open standards such as CityGML/CityJSON are essential to evolution of digital twin
development and integration. Translations between standards easier.
CityGML is an exchange format, not a client application (PDF) or storage format
(GeoDB). Need conversion to support end user applications – web / thin mobile clients:
GeoJSON, IMDF, KML.
Model / ETL based data harmonization provides direct control over semantics and
geometry structure. This allows the definition of transform models between
applications and standards, enabling support for updates and automation needed for
digital twins.

Conclusions
Continued collaboration and across BIM, 3D and GIS communities is essential. Data
models need not converge as they serve diﬀerent applications (3D GIS, CityGML3,
CityJSON, AR).
Approaches to enterprise automation for whole digital twin data life cycle:
generation, extraction, cleaning, conversion templates between open and industry
standards, validation, updates, publication, & cloud services - moving beyond a file
basis to services.
Emerging opportunities for indoor & underground mapping, AR, VR, 3D scanning.

Next Steps
Download
FME 2021.2
Download the
workspaces
from this
webinar
Apply it to your
own data to
manage changes
Free Trial | Upgrade
Contact us. We
would love to
chat with you
about anything
data
info@safe.com

Claim Your Community Badge
Get community badges for
watching webinars!
fme.ly/WebinarBadge Todayʼs Code: CFSBP

The Peak of Data Integration 2022 UC
August 24-26, 2022 Vancouver, Canada
Appy to present | Register now

FME World Tour 2022
March 22 - June 24 | 45+ Events
View the Tour Schedule & Register Now

Check out our upcoming
& on-demand webinars:
safe.com/webinars

Thank you!
Connect with us at info@safe.com
or in the Community
Download FME 2021.2 Free Trial | Upgrade
Connect with us for more FME
Please share
your feedback
with us through
the webinar
survey!

How to Easily Read and Write CityGML Data (Without Coding)

More Related Content

What's hot (20)

Similar to How to Easily Read and Write CityGML Data (Without Coding) (20)

More from Safe Software (20)

Recently uploaded (20)

How to Easily Read and Write CityGML Data (Without Coding)