Geo-Information

E4: 3D Visualization and 3D Animation

Warm Up

Common Types of 3D Data

3D data can appear in many formats and domains:

Point clouds 🟢 (from LiDAR, photogrammetry, laser scanning)
Meshes 🕸️ (triangular or polygonal surfaces from 3D modeling)
Voxel data 🧊 (3D pixels, often used in medical imaging or volumetric models)
3D building models 🏢 (e.g., CityGML, BIM data, CAD models)
DEM/DSM (Digital Elevation/Surface Models, raster grids with height values) {figure} ../images/ex4/3D-data-diff.ppm --- name: Illustration-of-3D-data --- The illustration of various of 3D data

2.5D(Pseudo 3D) vs. 3D Data

Aspect	2.5D (Pseudo-3D)	True 3D
`Geometry`	One z-value per (x, y)	Multiple z-values per (x, y) possible
`Representation`	Raster grid (DEM/DSM), extruded polygons	Point clouds, meshes, voxels, solids
`Complexity`	Relatively simple	More complex data structures
`Storage & Processing`	Lightweight, fast	Heavy, requires advanced computation
`Can model overlaps?`	❌ No	✅ Yes (bridges, tunnels, multi-floors)
`Main Use Cases`	Terrain, surface analysis, quick 3D view	Urban modeling, subsurface, indoor 3D

2.5D is enough when you only care about terrain or building heights.
True 3D is necessary when you need full volumetric analysis or overlapping structures.
In geodesy, we often start with 2.5D (e.g., DEMs) but increasingly move to full 3D for city models, infrastructure, and digital twins.
2.5D is also common in video games and VR.

How Do We Get 3D Data?

This is a huge topic that connects to many industries: Robotics, Autonomous Driving (AD), Gaming, Digital Design, Architecture, Geodesy, and more.
Broadly speaking, there are two ways: direct measurement (sensors record 3D geometry in real-world coordinates) and indirect reconstruction (3D is inferred from 2D observations or designed manually).

LiDAR Scanning (Airborne, Terrestrial, Mobile)

Captures dense point clouds with XYZ coordinates and intensity values.
Can be mounted on drones, aircraft, cars, or tripods.

Geodetic Surveying

GNSS / GPS surveying: directly measures 3D positions in global coordinates (WGS84, UTM, etc.).
Total station or laser theodolite: high-precision ground-based measurement for engineering, construction, and geodesy.

Photogrammetry (Multi-View Geometry)

Derives 3D geometry from overlapping images(aerial, drone, or satellite).
Large-scale projects: SRTM generated near-global elevation models.

3D Computer Vision

Structure from Motion (SfM): reconstructs camera positions + sparse point clouds from unordered photos, (COLMAP).
Multi-View Stereo(MVS): refines dense surface geometry given camera parameters, e.g. MVSNet
Deep learning approaches:
- VGGT (Vision Geometry Grounded Transformers): learns 3D scene geometry from multi-views.
- Neural rendering methods:
  - NeRF (Neural Radiance Fields) – photorealistic novel views.
  - 3D Gaussian Splatting – real-time 3D rendering.

Artificial Creation

Models created by artists or engineers in CAD, BIM, or 3D modeling tools (SketchUp, Blender, 3ds Max).
Common in gaming, movies, architectural visualization, digital twins.

```{admonition} With or Without World Coordinates?

Direct methods (LiDAR, GNSS, total station) → usually already georeferenced, so they can be integrated into GIS immediately.
Indirect methods (SfM, photogrammetry without GCPs, manual design) → typically start in relative coordinates only.
If no information about scale or world coordinates is available, the dataset cannot be directly placed in a real-world CRS. This is why we require Ground Control Points (GCPs) or another external reference source (e.g., GNSS measurements) to restore accurate georeferencing.
```

Task

Descriptions

In this exercise, you’ll move from 2D mapping to 3D visualization in ArcGIS Pro. You will set up and display spatial information in three dimensions, create realistic 3D symbols for buildings, street furniture, and monuments, capture GNSS points, and finish by producing a 3D animation and exporting it as a video. Detailed instructions in {download}Lesson 4 <../doc/Lesson 4.docx>

& You can Click here to look

Data

Data Students.gdb (buildings, roads)
Park-cadastral.tif
TUM.dae
Old_Pinakothek.dae
Monuments.gpx
Track.gpx

Overview

:class: dropdown
- [ ] **Prepare building layer and extrude into 3D**
  - assign heights and visualize buildings in three dimensions.  
- [ ] **Create and manage a Local Scene** 
  - set up a 3D workspace using a local UTM projection for accurate measurements.  
- [ ] **Digitize and symbolize new 3D point features** 
  - add trees, streetlights, benches, and landmarks from imagery.  
- [ ] **Classify and orient features** 
  - use attributes to control rotation, style, and size of symbols.  
- [ ] **Import and position custom 3D models** 
  - replace simple shapes with detailed building/monument models.  
- [ ] **Capture GNSS points and visualize in 3D** 
  - import GPS measurements and align them with the scene.  
- [ ] **Create a fly-through animation** 
  - set camera paths, transparency changes, and annotations.  
- [ ] **Export animation to video** 
  - produce a shareable MP4 file and submit.

1. Preparing and Extruding Buildings

Select buildings with level_num = 0 or NULL and assign default value 3.
Create a Local Scene and add Buildings and Roads.
Extrude buildings by $feature.level_num * 4 (meters per floor).

Change scene projection to ETRS 1989 UTM Zone 32N.
Explore 3D navigation tools and mouse controls.

3. Digitizing New 3D Features

Create point feature classes for trees, streetlights, benches, and landmarks with Z-values enabled.
Digitize features from cadastral plan and aerial imagery.
Place anchor points for TUM and Old Pinakothek.

4. 3D Symbolization

Apply built-in 3D marker symbols to point layers.
Classify park benches by Orient attribute and adjust rotation.
Replace extruded landmark buildings with custom 3D models and position them correctly.

5. Capturing & Displaying GNSS Points

Use GPS device to record monument locations (20 averaged points each).
Convert GPX files to shapefiles using GPS Utility.
Define projection as WGS 1984 and symbolize monuments with custom 3D models.

6. Creating a 3D Animation

Plan keyframes for viewpoints and transitions.
Adjust transparency, add titles, and set duration.
Export animation as MP4 (YouTube preset) and submit.

Advance Task

Automate the workflow Try scripting repetitive steps such as updating attributes, extruding buildings, or converting data. ```{code} python
Update Building Heights from CSV

import arcpy building_fc = “path/to/Buildings” csv_table = “path/to/building_heights.csv” arcpy.management.AddJoin(building_fc, “building_id”, csv_table, “building_id”) expression = “!building_heights.levels! * 4” arcpy.management.CalculateField(building_fc, “extrude_height”, expression, “PYTHON3”)

Convert GPX to Shapefile (Python + Geopandas)

import geopandas as gpd gpx_file = “monuments.gpx” points = gpd.read_file(gpx_file, layer=’waypoints’) points.to_file(“monuments.shp”, driver=”ESRI Shapefile”) ```

Exploring Heights Across Coordinate Systems Elevation (Z-values) is not as straightforward as it seems. The meaning and interpretation of “height” depends heavily on the coordinate system in use. You can investigate how height values described on the different spatial reference system.
Visualize Real Flight Trajectories in 3D
- Explore real-world aviation data, download a flight track, visualize it as a 3D path in Google Earth or ArcGIS Pro, and see how these data is organized.
- Download and process Real Flight Data from flight websites like:
- Visualize in Google Earth
  - Google Earth can directly open KML or GPX files with 3D elevation data.