3D Digitalization of an 8,000 m² Landscape Architecture Training Base

A landscape garden is more than a practical site for landscape design. It is also an important teaching and training resource for landscape architecture students. However, making the most of this living classroom presents several challenges:

• How can the spatial layout, architectural elements, and landscape details of the entire garden be accurately documented?

• How can students access, explore, and measure the site without being restricted by time or location?

• How can captured 3D data be transformed into practical resources that support both teaching and site management?

Recently, TBNET successfully completed a digitalization project for the Landscape Architecture Training Base at Hubei Urban Construction Vocational and Technological College.

The project adopted a lightweight solution integrating handheld 3D scanning, 3D Gaussian Splatting reconstruction, and a no-code visualization platform. Beyond creating a digital record of the garden, the solution transformed 3D data into accessible digital resources that teachers, students, and site managers can use directly, establishing a complete workflow from the physical site to 3D data and practical educational applications.

Looking back at the project, four key challenges were addressed.

Established in 2016, the Landscape Architecture Training Base at Hubei Urban Construction Vocational and Technological College covers approximately 8,000 m². Jointly developed, managed, and shared by the college and industry partners, the base serves as a practical training facility for industry-education integration.

Incorporating the concepts of sponge city development and urban regeneration, the site brings together various elements of traditional Chinese garden design. Pavilions, terraces, covered corridors, waterscapes, and landscaped areas are distributed throughout the site.

However, its location within a developed urban area and its complex spatial environment presented several challenges for data acquisition:

• Restricted airspace: Strict airspace regulations prevented the use of drones for aerial data collection.

• Weak satellite signals: GNSS signals were limited inside pavilions and covered corridors, as well as in shaded areas beneath trees.

• Complex spatial features: Intricate architectural details and scattered landscape elements made complete coverage difficult using conventional surveying methods.

To address these challenges, the project used the Lixel L2 Pro handheld real-time 3D mapping device. The technical team walked continuously along the garden paths, completing a full-site scan in approximately 40 minutes with a single operator. The system directly captured georeferenced, true-color point cloud data.

Compared with traditional surveying methods that require multiple instrument setups and separate measurements across different areas, handheld scanning is better suited to gardens with complex internal spaces and widely distributed landscape features. It also significantly reduces the time and manpower required for field data collection.

After data acquisition, the project processed the collected data according to the different requirements of surveying and engineering applications and realistic 3D visualization, producing two categories of deliverables.

1) Point Cloud Data for Surveying and Engineering Applications

Using LixelStudio, the team performed dynamic noise filtering, point cloud classification, vector drawing, and format conversion. The processed data was exported in commonly used formats such as .las, providing a reliable data foundation for surveying, engineering quantity calculations, and subsequent modeling.

2) Gaussian Splatting Models for Realistic 3D Visualization

The collected data was also processed using Lixel CyberColor (LCC) to generate 3D Gaussian Splatting models. These realistic 3D scenes provide the foundation for interactive navigation, landscape visualization, and integration into the digital platform.

The two types of deliverables serve different purposes. By applying separate processing workflows to the same dataset, the project extended the value of field data collection beyond acquisition, supporting both engineering applications and immersive 3D scene visualization.

The reconstructed Gaussian Splatting models were integrated into the HDSL 3D Visualization and Micro-Application Platform.

Leveraging the platform's 3D rendering engine and no-code development capabilities, users can create interactive applications without writing code. Functions such as hotspot annotation, educational content creation, and link publishing can all be configured directly through the platform.

This platform-based approach transforms the project deliverables from a one-time handover into digital assets that the college can independently maintain and continuously update.

The training base can now support five major application scenarios for teaching, management, and presentation.

01. Overall Site Overview

Users can access the entire garden through a shared link, viewing the overall layout from above or exploring individual courtyards at ground level. The site overview is integrated with the 3D scene, providing a more intuitive way to understand the training base.

During visits and presentations, the scene can be accessed directly through a web link without installing dedicated software.

02. Interactive Hotspot Annotation

Hotspots can be added throughout the 3D scene and linked to supporting materials, including text, images, audio, and video.

This allows knowledge about architectural elements, plant arrangements, and traditional landscaping techniques to be associated with their actual locations. Students can explore these details visually and access relevant learning materials whenever needed, making teaching content more flexible and reusable.

03. Guided Scene Presentations

By setting predefined navigation paths, users can generate guided walkthrough videos with a single click.

Camera movements and viewing directions can be arranged according to presentation requirements, providing a consistent guided tour experience for teaching activities, visitor reception, and promotional presentations.

04. Spatial Measurement

Using the real-world coordinates embedded in the 3D data, users can measure individual points, distances, and areas.

CAD drawings and DEM models can also be overlaid for comparative analysis. These functions support practical training activities such as traditional architecture surveying, garden pathway quantity calculations, and pavement area measurements.

05. Free Exploration

The interactive 3D environment allows users to closely examine architectural and landscape details, including roof tiles, dougong bracket systems, brick joints, and waterside stone features.

This provides valuable visual resources for courses covering traditional architecture appreciation, architectural component identification, and plant observation.

For site managers, the platform also enables remote inspection of existing site conditions. By comparing datasets captured at different times, they can monitor changes across the garden and its facilities.

The value of garden digitalization goes beyond completing a data acquisition task or creating a 3D model. What matters more is whether these digital assets can become part of everyday teaching, practical training, and site management.

Through lightweight data acquisition, categorized processing, and platform-based applications, this project has taken 3D data beyond the delivery stage and brought it into practical use.

For teaching and practical training, the 3D environment extends the classroom beyond the physical site. Students can closely examine traditional architectural components, garden paving, and plant arrangements. They can use the digital scene for traditional architecture appreciation, component identification, and plant observation, as well as practical exercises such as architectural surveying and pathway quantity calculations.

For site management, 3D data provides an effective way to inspect and compare site conditions. Managers can remotely access the current state of the training base and compare datasets from different periods to identify changes in facilities, vegetation, and the surrounding environment, providing useful references for routine management.

For presentations and visitor reception, interactive 3D scenes offer a more intuitive way to introduce the training base. Through shared links or predefined virtual tours, the college can use the digital environment for campus visits, admissions promotion, and visitor presentations, allowing people who cannot visit in person to explore the garden's layout and landscape features.

By connecting field data acquisition, data processing, platform development, and practical applications, the project brings previously separate technical processes into one integrated workflow. As a result, 3D data is no longer limited to storage and visualization but can directly support specific teaching and management needs.

For landscape architecture education, these digital resources provide more than a record of spatial layouts and landscape details. They also create accessible learning materials for architectural component identification, traditional building surveying, and plant observation.

This project demonstrates how a landscape architecture training base can move beyond traditional digitalization approaches, which often involve high investment, lengthy workflows, and technical barriers, toward a digital asset that the institution can independently operate and maintain.

As field data acquisition becomes more convenient, 3D reconstruction more automated, and application development more accessible, cultural spaces such as traditional Chinese gardens can be digitally preserved with greater efficiency and lower operational costs.

Most importantly, 3D models no longer remain as files stored on a computer. They become practical digital resources that enter the classroom, support everyday management, and continue to deliver value over time.