Over a sustainable period of time, the life cycle of buildings has been extended in the process of continuous repair and maintenance, and in addition to the comprehensive value of history, craftsmanship, aesthetics, and cultural diversity, these “solidified epics” have now added a high-tech dimension.

On April 26, the “Smart Yingxian Wooden Tower” program, jointly developed by Lenovo Group and the School of Architecture of Tsinghua University and donated to the Yingxian Muta Scenic Area, debuted. The thousand-year-old wooden spectacle collided with the latest spatial computing AIGC combination technology to create a unique spark.

Yingxian Wooden Pagoda is the only oldest and tallest wooden pavement currently existing in the world. It has been gradually damaged by earthquakes, wind, rain, war and other factors in the past thousand years. Its internal structure has been deformed, and its carrying capacity has been greatly reduced, and visitors are no longer able to climb and visit the tower.

AIGC refers to content automatically created and generated by AI (AI Generated Content), that is, AI receives task instructions from the recipient and automatically generates images, videos, audio, etc. through processing the natural language of the person. ChatGPT and Midjourney are all representative works of technology.

“In the past, our cultural relics protection was modeled by hand. The industry first proposed using neural radiation field technology for 3D reconstruction. The world's first paper was born in 2020. Lenovo quickly introduced neural radiation field-related technology and worked with Tsinghua University to explore how to model wooden towers using the latest spatial calculations.” Mao Shijie, vice president of the Lenovo Group and director of the Lenovo Shanghai Research Institute, told the Wall Street story.

In the digital world, a “wooden tower twin” is one of the benchmark works of Lenovo's “New IT, New Heritage” strategy, and it has also become a new experiment in technology to enable sustainable development.

Digital restoration: 10 hours, 15 million sheets

The book “Yingxian Wooden Pagoda” has compiled and counted the recorded history of wooden pagoda repairs. Since the wooden tower was built until 1949, it has undergone six major repairs, with an average of once every 150 years.

Since the restoration work of the wooden pagoda in Yingxian County was officially put on the agenda in the mid-90s, the wooden tower protection and restoration plan has become the focus of ancient architecture research. The various repair plans include partial repairs, overhauls, full support plans, and dismantling repairs, etc., which are all mixed.

Liu Chang, director of the Institute of Architectural History and Cultural Heritage and Building Protection at the School of Architecture of Tsinghua University, confessed to Wall Street that there are three major thresholds for wooden tower protection in Yingxian County. “The first is extensive preliminary preparation, capital investment, team building, and project setting. The second is understanding the main body, which requires more ingenuity, and the third is the extensive collection of new technology.”

This time, the “Smart Yingxian Wooden Tower” provided a breakthrough solution under the AI discourse system.

“Every time we've been here in the past two years, it's different from traditional restoration work. We used to be model engineers. We used drones, cameras, and radar cameras to collect images using various new non-destructive detection methods.” Mao Shijie told Wall Street News that during the research and development process, the project used more than 15 million sheets of basic materials to complete nearly 60,000 lines of Unity programming, and the rendering model reached 4.2G.

If traditional techniques were used to depict and trace, the basic analytical material for processing massive photographs of wooden towers from various angles would be extremely costly in human and material resources, and Lenovo innovatively used a combination of artificial intelligence (AI), neural radiation field (NeRF) technology, and extended reality (XR) technology to greatly improve work efficiency.

Mao Shijie explained to Wall Street News that the world's first paper on AIGC technology appeared in 2020, with the goal of building a digital world. Since then, Lenovo has invested in the development of this technology.

“The deep hybrid rendering model generated in the middle cannot actually be used directly. It is essentially different from the 3D model constructed by the traditional model. It is a neural network model, so there are many technical problems to be solved. It needs to be mixed and rendered with the traditional model, including splicing these models with each other, etc. These are all problems that need to be solved in addition to the algorithm itself.” Mao Shijie said.

According to reports, after using a radar camera, RGB camera, and 360Camera to collect data inside the wooden tower, all remaining data was calculated using artificial intelligence algorithms, and it took 10 hours to restore it to a 3D model.

The School of Architecture at Tsinghua University is responsible for modeling the wooden structure inside the wooden tower and creating a database of wooden tower parameters through structural parameter research; at the same time, the research results are transformed into wooden tower displays and popular science content to create scripts for virtual experiences.

In the end, the “Smart Yingxian Wooden Pagoda” experience application recreated 5 major exploration scenes on the 1st to 5th floors of the replicated wooden pagoda, covering 7 major interactive designs such as talking to Liang Sicheng and picking up AR props, as well as 11 large-scale animation effects, including Buddha statues from the Tibetan Sutras and Buddha ceremonies.

Information Inheritance: The “Fourth Stage” of Technological Empowerment

According to an analysis by an architectural aesthetic researcher, the “Venice Charter” affirms the identity of architecture as a “witness of history and times”. In order to ensure the current status and safety of the building, it promotes not making any changes to the architectural form, but instead allows it to be rooted in history like a specimen, passed on to future generations and become an eternal asset for all mankind. However, this “frozen” inheritance method further reflects the protection perspective of Western architecture with brick and concrete as the main materials. The “Narahara Truth Documents” indicate that architectural heritage is a bearer of historical and cultural information. For overall information other than “specimens” Attention is more applicable to the context of Oriental architectural protection, which mainly focuses on wooden architecture systems.

Empowering wisdom with AI, Yingxian Muta is a new digital protection paradigm for “new IT heritage” built by Lenovo based on a new IT technology architecture of “end - edge - cloud - network - intelligence”. It extends the entire life cycle of architectural heritage information in the digital world.

Wall Street News learned that currently Lenovo Group's spatial computing AIGC technology is divided into three core stages: the first stage is to use various acquisition devices such as the latest generation drone equipment and 360° cameras to surround the wooden tower to complete the collection of thousands of image data inside and outside the wooden tower, and accumulate a certain amount of basic material;

The second stage is to use AI technology to analyze and process the collected basic material data, train its intelligent perception and understanding of 3D spatial information through deep learning and neural networks, construct a highly simulated 3D scene, and quickly and accurately reproduce the details of the building;

The third stage uses XR technology to seamlessly connect the generated 3D model with the real world or virtual world, and to achieve flexible application and interaction of the 3D model in different scenarios through integration and reconstruction of fragmented information.

And the technology expectations of industry insiders may mark the “fourth phase” of future empowerment.

“For digital technology to support cultural relics protection, the most important thing is actually to solve the role of information extraction and provide basic support for research. Take the wooden tower as an example. Although much of the information on the wooden tower has gone through continuous manual mapping, 3D scanning, and total station mapping, some information is unknown.” Wang Xiaolong, deputy director of the Shanxi Ancient Architecture and Color Plastic Mural Conservation Research Institute, told Wall Street News that as a researcher, he hopes to improve the true understanding of wooden towers in the future, whether through big data AI algorithms or more sophisticated surveying and mapping technology.

“Including understanding of its scale, perhaps even moving from the perception of the existing scale to the scale of its original design, the perception of the different stages of scale, and the perception of its disease, such as how large is the decay area of wood? How deep is the decay? If digital technology can directly extract this information for us, subsequent protection efforts will be targeted.” He analyzed it this way.

In Liu Chang's imagination, it is hoped that in the future, AIGC technology can “paint the bones” of wooden towers. “Cultural relics protection professionals need to see not only the appearance of a person's face, but also the internal bones, and even the distribution of nerves and muscles within. Mathematics can not only achieve AIGC, but also act as a bridge between different data. Therefore, from the perspective of cultural protection professionals, AIGC is now painting skin, but in the future, I hope it can draw flesh and bone, and I hope to link with other data media.”

According to reports, on the basis of existing data files, the Yingxian Muta project has also added a number of new tests in conjunction with AI technology. For example, AI and AR technology are used to assist site surveys, view basic building conditions and valid data information in a virtual space through remote operation, and assist in completing related research.