From Thousands to Seamless: The Power of Virtual Tilesets in Thailand's 3D Landscape

Managing national-scale data, such as 41 million building items for Thailand, requires flexibility for continuous data updates and peak performance without sacrificing efficiency.

Since the initial monolithic approach was non-scalable and costly, the solution was Data Partitioning: breaking the massive dataset into smaller, manageable chunks. We used the powerful Google S2 grid system to organize the country's geometry data into defined processing cells. This structure, combined with 3D Tiling and the Virtual Tileset concept, creates flexible 3D collections. This ensures that when a single building is updated, we only re-process its tiny corresponding cell, guaranteeing massive time savings, high efficiency, and significantly reduced operational costs.

Transcript

1. From Thousands to Seamless: The Power of Virtual Tilesets in Thailand's 3D Landscape Siriwat Suttipanyo | Siriya Saenkhom-or Vallaris Maps, i-bitz company limited FOSS4G Auckland 2025, 17 - 23 November 2025
2. Vallaris Maps We are Modern Geospatial Solution Ref: https://unsplash.com/photos/ucYWe5mzTMU Vallaris Maps is a geomatics platform that enables easy map creation with high flexibility in usage and delivers data services according to standards certified by The Open Geospatial Consortium (OGC®) – the first in Asia. It offers diverse options for map utilization on various devices including workstation, smartphones and tablets. It serves multiple industries and businesses.
3. Overview https://3d.gistda.or.th The project's goal was to engineer a web application capable of managing and visualizing Thailand's entire national building dataset. The core requirements were to achieve high-performance rendering of this massive data while maintaining a simple and flexible management interface.
4. Sources • GISTDA • Google Open Buildings • Microsoft Building Footprint Building Data of feature items are stored on Vallaris Maps Platform ~41M
5. 01 02 03 Our Goals Visualize building footprints as 3D Tiles accurately positioned on a real-world Digital Elevation Model (DEM) Provide comprehensive data management allowing users to create, delete, and edit the geometry or properties of any building Ensure high performance for both visualization and data processing, even with a massive, national-scale dataset
6. 01 02 03 04 Processing Scalability The approach of processing all 41M features at once failed and was not a scalable solution. Performance The processing time for a single, monolithic build was impractically long and would never finish. Data Maintainability The "Full Generate" method was inflexible, as it required re-processing all features just to update a single building. Cost & Resource Optimization Processing all 41 million features in a single batch was was prohibitively costly and resource-intensive. The Challenges
7. Data Partitioning using Google S2 Grid Merge multiple tilesets using Cesium 3D Tiles Tools 3D Tiles Generation using Mago3DTiler Manage with Virtual Tileset Techniques Implemented
8. Why We Use S2 Global Grid System It provides a single, consistent grid for the entire globe. Data Partitioning It allowed you to break down 41 million features into smaller, manageable, geographically-based chunks. https://en.wikipedia.org/wiki/Discrete_global_grid
9. Why Level 9 Appropriate Cell Granularity Level 9 provided the optimal cell dimensions for effectively grouping the building footprint features. Optimized Processing Batches This level successfully partitioned the 41M-feature dataset into many small, manageable workloads is the maximum number of features within a single S2 Level 9 grid cell (Bangkok-Nonthaburi area) ~950K 1535 is the number of S2 grid level 9 that cover Thailand area
12. mago 3DTiler OGC 3DTiles generator An open source-based Effortlessly convert an array of 3D formats, including 3DS, OBJ, FBX, Collada DAE, GlTF, GLB , IFC and more. Multi-Format Mastery Turn 2D geospatial data (GeoJSON) into detailed 3D extrusion models with GeoTIFF DEM to set 3D model clampToGround 2D to 3D Extrusion https://github.com/Gaia3D/mago-3d-tiler
13. Right now we have 1535 tileset.json files that ready to render on web
15. Based on the 3D Tiles root-children hierarchical relationship and external tilesets concept When editing a specific 'child' cell, we only regenerate that child's 3D Tiles and then re-calculate the 'root' tileset. which consists of links to all the 'child' 3D Tiles from each S2 grid cell. Virtual Tileset Concept Acts as the single 'root' (starting point) Localized Updates
16. Cesium 3D Tiles Tools Run commands in a terminal to perform specific, powerful tasks A command-line toolkit For converting, optimizing, processing, and analyzing 3D Tiles data Tools and utilities Merge multiple tilesets into a single tileset JSON file mergeJson https://github.com/CesiumGS/3d-tiles-tools
17. Initial 3D tileset Visible-Only Rendering
19. The Processing Results is the time took to process the virtual tiles children merging (validate + mergeJson + save) ~19secs is the time took to process 3D Tile generation for Bangkok- Nonthaburi area S2 grid cell ~15mins The service is running on a Kubernetes cluster with shared resources. Worker resource usage - CPU: 3 Core - RAM: 20 GB Service resource usage - CPU: 0.4 Core - RAM : 70 MB 3D Tile Generation Virtual Tileset Merging
20. Key Learnings Monolithic processing is impractical Processing a massive dataset in a single batch is inflexible and will fail to scale. Partitioning enables modularity Using the S2 Grid allows for managing tiles independently, so a local data change only requires a small, local update. The Virtual Tileset is the "glue" This is the key component that unifies the partitioned architecture, making all the independent tiles appear as a single, seamless layer to the user. 01 02 03
21. Next Steps Complex 3D Transformation Investigating methods to efficiently manage transformations (translation, rotation, or scaling) on individual child tilesets and correctly reflect those changes within the main virtual tileset.
22. If you have any questions, feel free to contact us any time on mobile phone or contact us by email. We will get back to you as soon as we can. i-bitz company limited 1371 Suite 10 Capital Mansion, Phahonyothin Rd, Phayathai, Bangkok, 10400 Thailand Office Hours Monday-Friday 09.00-17.00 Get In Touch (+66) 2278 7913 More Information www.i-bitz.co.th sales@i-bitz.co.th Thank You for Joining This Session