Sustainable Column | Digital Twin: Lenovo's "Wooden Tower" Restoration Technique

In the perpetual time, the lifecycle of buildings is extended through continuous repair and maintenance, and these "frozen epics" now add a new dimension of high-tech in addition to their comprehensive value of history, craftsmanship, aesthetics, and cultural diversity.

On April 26th, the "Smart Yingxian Wooden Pagoda" project developed and donated by Lenovo Group in collaboration with the School of Architecture at Tsinghua University made its first appearance for use in the Yingxian Wooden Pagoda scenic area. The ancient wooden structure, standing for a thousand years, collided with the latest spatial computing AIGC technology, sparking a unique innovation.

The Yingxian Wooden Pagoda is the oldest and tallest existing pure wooden structure tower-style building in the world. Over its nearly thousand-year history, it has accumulated damage from factors such as earthquakes, wind, rain, and wars, causing deformation in its internal structure and a significant reduction in load-bearing capacity, making it inaccessible for visitors to climb and visit.

AIGC refers to AI Generated Content, which is content created automatically by AI. AI receives task instructions from humans and generates images, videos, audio, etc., through processing natural language. ChatGPT and Midjourney are representative technologies in this field.

"In the past, our cultural relic protection was all done manually. For the first time in the industry, we proposed using neural radiation field technology for 3D reconstruction. The world's first paper was published in 2020. Lenovo quickly introduced related neural radiation field technology and, together with Tsinghua University, explored how to model the wooden pagoda using the latest spatial computing," said Mao Shijie, Vice President of Lenovo Group and Dean of Lenovo Shanghai Research Institute, to Wall Street News.

In the digital world, a "Wooden Pagoda Twin" is one of the flagship works of Lenovo's "New IT, New Cultural Heritage" strategy, and it has become a new attempt to empower sustainable development with technology.

Digital Restoration: 10 hours, 15 million faces

The book "Yingxian Wooden Pagoda" has compiled and analyzed the recorded history of repairs to the wooden pagoda. From its construction to 1949, it has undergone six major repairs, averaging one major repair every 150 years.

Since the mid-1990s, the restoration of the Yingxian Wooden Pagoda has officially become a priority, making protection and restoration plans for the wooden pagoda a focus of ancient architectural research. Various repair plans include partial repairs, large-scale dismantling and repair, full support plans, and dismantling repairs, among others.

Liu Chang, Director of the Institute of Architectural History and Heritage Conservation at the School of Architecture, Tsinghua University, admitted to Wall Street News that there are three major challenges in the protection of the Yingxian Wooden Pagoda: "First, a large amount of preliminary preparation is required, including funding, team building, and topic setting, which is very difficult. Second, understanding the object itself requires more intellectual input. Third, collecting new technologies widely."

The "Smart Yingxian Wooden Pagoda" project provides a breakthrough solution under an AI discourse system.

"In the past two years, every time we came here, it was different from traditional repair work. Previously, model engineers came, but now we use drones, cameras, radar cameras, and various new non-destructive detection methods to collect images," Mao Shijie told Wall Street News. During the research and development process, the project accumulated the use of basic materials exceeding 15 million faces, completed nearly 60,000 lines of Unity programming, and rendered models up to 4.2G If traditional techniques are used to depict and copy, processing a large number of photos of the wooden pagoda from various angles will consume huge manpower and resources. However, Lenovo innovatively combines artificial intelligence (AI), Neural Radiance Fields (NeRF) technology, and Extended Reality (XR) technology to greatly improve work efficiency.

Mao Shijie explained to Wall Street News that the first global paper on AIGC technology appeared in 2020, with the goal of building a digital world. After that, Lenovo began investing in the research and development of this technology.

"The model generated in the middle of the deep mixed rendering process cannot be directly used. It is fundamentally different from traditionally constructed 3D models, as it is a neural network model. Therefore, there are many technical issues to be solved, such as blending it with traditional models, stitching these models together, and so on. These are all problems that need to be addressed in addition to the algorithm itself," Mao Shijie said.

It is reported that after collecting data inside the wooden pagoda using radar cameras, RGB cameras, and 360 cameras, all remaining data is calculated by artificial intelligence algorithms, and a three-dimensional model is restored in just 10 hours.

Tsinghua University's School of Architecture is responsible for modeling the wooden structure inside the pagoda and creating a database of pagoda parameters through structural parameter research. At the same time, research results are transformed into exhibition and popular science content for the pagoda, creating scripts for virtual experiences.

The "Intelligent Yingxian Wooden Pagoda" experience application ultimately restores and replicates five major exploration scenes on the first to fifth floors of the pagoda, covering 7 major interactive designs such as conversing with Liang Sicheng, picking up AR props, and 11 large-scale animated special effects scenes such as Buddha statues and worshipping scenes.

Information Inheritance: The "Fourth Stage" of Technological Empowerment

An architectural aesthetics researcher analyzed that the "Venice Charter" affirmed the identity of architecture as a "witness to history and the times." To ensure the current status and safety of buildings, it advocates not making arbitrary changes to architectural forms, but rather letting them be rooted in history like specimens, passed down to future generations, and becoming eternal wealth for all humanity. However, this "freezing" method of inheritance more reflects the protection perspective of Western architecture mainly made of bricks and concrete. The "Nara Document on Authenticity" points out that architectural heritage carries historical and cultural information, enhancing the focus on overall information beyond the "specimen," which is more suitable for the context of protecting Eastern architecture mainly based on wooden frame systems.

Empowering the Intelligent Yingxian Wooden Pagoda with AI is a new paradigm of digital protection called "New IT New Cultural Heritage" built by Lenovo based on the "End-Edge-Cloud-Network-Intelligence" new IT technology architecture, extending the lifecycle of overall information of architectural heritage in the digital world.

Wall Street News has learned that Lenovo Group's Spatial Computing AIGC technology is divided into three core stages: the first stage involves using the latest generation of drones and various collection devices such as 360° cameras to scan and photograph the pagoda from all angles, capturing thousands of images inside and outside the pagoda to accumulate a certain amount of basic materials The second phase involves using AI technology to analyze and process collected basic material data, and through deep learning and neural network training, intelligently perceive and understand three-dimensional spatial information, constructing highly realistic three-dimensional scenes, quickly and accurately reproducing the details of buildings;

The third phase involves using XR technology to seamlessly connect the generated three-dimensional models with the real world or virtual world, integrating and reconstructing fragmented information to achieve flexible application and interaction of three-dimensional models in different scenarios.

Industry insiders' technological expectations may signal the "fourth stage" of future empowerment.

"When it comes to supporting cultural relic protection with digital technology, the most important thing is actually solving information extraction, providing a foundation for research. Take wooden pagodas as an example, although a lot of information about wooden pagodas has been continuously collected through manual surveys, three-dimensional scanning, and total station surveys, there is still some unknown information." Wang Xiaolong, Deputy Director of the Institute of Ancient Architecture and Colored Sculpture Protection in Shanxi Province, told Wall Street News, "As researchers, we hope that in the future, whether through big data AI algorithms or more refined surveying techniques, we can enhance our true understanding of wooden pagodas."

"This includes understanding its scale, possibly even pushing towards its original design scale from the existing scale, as well as understanding the scale at different stages, and understanding its diseases, such as how large is the area of wood decay? How deep is the decay? If digital technology can directly extract this information for us, the subsequent protection work will be more targeted." He analyzed.

In Liu Chang's vision, he hopes that in the future, AIGC technology can "draw bones" for wooden pagodas. "In the field of cultural relic protection, professionals not only need to see a person's facial appearance, but also need to see the internal bones, and even the distribution of nerves and muscles inside. Mathematics can not only achieve AIGC, but also act as a bridge between different data. Therefore, in the eyes of cultural relic protection professionals, AIGC is currently drawing the skin, but in the future, we hope it can draw flesh, draw bones, and connect with other data media."

It is reported that based on existing data archives, the Xing County Wooden Pagoda project has also added multiple attempts combining AI technology. For example, using AI and AR technology to assist on-site surveys, remotely viewing basic building information and effective data in virtual space through remote operations, and assisting in completing related research, etc.