Overview of Common 3D GIS Data Models and Their Applications

Introduction

In 3D visualization, CIM, digital twin, and metaverse concepts, the core technical support relies on 3D GIS as a foundational base, which uses various 3D models as data carriers.

1. Tilted Photography Measurement Data Model

Tilted photography is a representative 3D model in GIS, offering high production efficiency and strong realism, but it essentially remains a "skin" lacking semantic information, limiting spatial analysis and queries, and its visualization suffers from fixed lighting and smoothness, often requiring secondary editing or combination with other data for fine-grained display.

2. Single-Body (Object) Modeling of Tilted Photography

To overcome the shortcomings of tilted photography in GIS applications, single-body techniques aim to give the model feature attributes enabling spatial queries and analysis. Four main approaches are described:

2.1 Physical Cut Single-Body Modeling

Good features such as buildings are clipped from the tilted model using building vector data to extract continuous triangular faces within the vector range.

2.2 Vector套绘 and Drawing Single-Body Modeling

Instead of physical cut, vector geometry is used in rendering: GPU tests whether vertices lie inside the vector area, highlighting them via color or shadow volume techniques; the vector’s attributes can serve as the single-body’s attributes. An alternative is to draw a transparent colored geometry that hides/unhides on selection.

2.3 ID Single-Body Modeling

Each region to be single-bodied receives a unique ID; during rendering an ID texture is sampled to highlight selected vertices. This method needs preprocessing from both data and rendering sides, has minimal performance impact, and is used in Cesium 3D Tiles via BATCHID, but requires complex preprocessing.

2.4 Model Reconstruction Single-Body Modeling

Manual reconstruction based on stereoscopic photogrammetry captures sharp, vertically aligned models, especially for buildings and infrastructure, improving visualization and enabling true single-body separation, though it demands professional expertise and software.

2.5 Scheme Comparison

No single approach is optimal; the choice depends on specific project needs, as summarized in a comparison table.

3. OSM Building Models

Using OpenStreetMap vector data, building footprints are extruded with height attributes to generate low‑cost city‑scale 3D models; textures, materials, and ancillary details (roofs, doors, foundations) can be added to improve realism, making OSM models suitable for stylized or non‑photorealistic urban visualization.

4. CAD Layered and Unit Models

CAD floor‑plan or unit drawings are extruded to create building models with internal layer‑and‑unit structure, which can be linked to door‑way attributes for fine‑grained management; when lacking georeference, the CAD data is aligned with building footprint vectors.

5. Building Information Model (BIM)

BIM provides detailed microscopic data that can complement GIS’s macroscopic geographic context, enabling applications such as indoor navigation and asset management; challenges include high acquisition cost, diverse proprietary formats, large data size affecting visualization performance, lack of geographic information, and the need to assign materials for proper visualization.

6. Manual Modeling

Hand‑crafted 3D models deliver the highest visual fidelity, especially for lighting and shadows, but are costly and time‑consuming; a practical strategy combines low‑cost models (e.g., OSM) for broad areas with manual models for key showcase zones to balance cost and visual quality.

7. Conclusion

Beyond model types, projects must address varied formats, organization, and lightweighting for scheduling and rendering; platforms like the JingShi CIM system provide mature solutions that have been widely applied in urban‑management projects using 3D GIS technology.