Xinjiang Today: How one researcher brings the ancient murals of Xinjiang's Kizil Caves back to life

BEIJING, March 15, 2026 /PRNewswire/ -- The Kizil Caves, located in Baicheng County, Aksu Prefecture in west-central Xinjiang, are among the earliest Buddhist cave complexes in China. Over a thousand years ago, Buddhism was introduced from ancient India to the western part of China. The earliest center of Buddhism in the Western Regions (now Xinjiang) was the ancient kingdom of Qiuci, today's Kuche (Kuqa).

Renowned for their exquisite murals, the Kizil Caves house a central collection of murals representative of the ancient Qiuci State's Buddhist art. Of the 349 caves, 107 contain murals spanning nearly 5,000 square meters. The Kizil Caves are one of the Four Great Grottes of China, along with the Dunhuang Mogao Grottoes in the northwest, the Yungang Grottoes in the north and the Longmen Grottoes in central China, and were recognized as a UNESCO World Heritage Site in 2014.

In the late 19th and early 20th centuries, many of the murals were plundered by foreign expeditioners and scattered across the world. Zhao Li, a research fellow at the Kizil Caves Research Institute in Xinjiang, has dedicated her life to restoring them, traveling overseas to retrieve lost fragments and recreating them digitally.

Zhao spoke to Xinjiang Today about the evolution of research on the murals of Kizil Caves and international collaboration to preserve them. This is an edited excerpt of the interview:

Xinjiang Today: How did you get started on researching the murals at Kizil Caves? How has international research on them developed over the past three decades?

I started research on this back in 1998, publishing papers and securing research projects from the National Cultural Heritage Administration. I also had exchanges with international experts in the area.

Foreign scholars have conducted extensive research on the cultural relics and murals of Kizil Caves taken away by expedition teams in the early 20th century. The last such team was the German expedition team that arrived in around 1913, and they took away many murals, as well as some statues and books. They published some reports in the 1920s and their publications remain essential references for us.

In the late 1970s, Su Bai, a professor with Peking University, went to the Kizil Caves with four students. They stayed in the area for three months, dating and periodizing the caves, and Chinese scholars gradually took the lead in the research.

By the 1990s, local experts such as Huo Xuchu, a senior researcher at the Kizil Caves Research Institute, and Jia Yingyi, a research fellow at the Xinjiang Museum, also became involved in the research.

I was a visiting scholar at the Museum of Asian Art in Berlin, Germany, from 2012 to 2013. I also visited the Hermitage Museum in Saint Petersburg, Russia, and the Guimet Museum in Paris, France. I feel our research is backed by stronger support and more resources by our own country.

What work have you done in cultural relic surveys and in-situ verification? What challenges did you encounter?

There are 486 Kizil Cave mural fragments overseas, according to our investigation. We have conducted in-situ verification at the caves to determine which cave and specific location each mural originally was taken away from.

The verification is a laborious process. We need to determine which cluster the mural belonged to, and then pinpoint the exact cave. Large murals are easy to identify, but the location of smaller ones, especially fragments, are difficult to pinpoint. The numbering system used by the German expedition team for the caves was unscientific, and many cave records were incorrect.

For example, the German team recorded a famous mural, The Royal Couple, as from Cave 13. However, we could not find its position in Cave 13. Then based on the black-and-white photos taken by foreign expeditioners, we later found that it was actually from Cave 171 in the eastern section.

Many cave walls were dilapidated when the foreign expeditions arrived. After they removed the murals, the walls collapsed, making it hard for (later) researchers to locate their original positions.

However, some caves are intact. Despite frequent earthquakes, the central pillars in the caves are quite stable. Without this architectural design, many caves might have collapsed, and we would not have been able to see them today. By 2020, we had identified the original positions of over 420 murals and restored them (by reproducing them via digital technology and then putting them back to their original locations). The original position of about 40 fragments still cannot be identified.

What has been your experience collaborating with foreign museums and scholars on surveys and research?

The Museum of Asian Art in Berlin has the largest collection of Qiuci murals, followed by the Hermitage Museum. Eleven institutions in the United States also house several Kizil Cave murals.

Huo Xuchu, my mentor, guided me in researching overseas murals. To conduct research, you have to be thoroughly familiar with the caves. While searching for the original position of a mural, I would think about it even during meals or while sleeping. I memorized exactly which murals were removed, which have been located, and which gaps remain unfilled.

In 2022, Huo and I attended international conferences in Germany and viewed many murals collected in the museum. That strengthened my resolve to go to Berlin again for research.

It took me 10 years to gain the opportunity to study in Berlin. While I was doing mural research at the Museum of Asian Art as a visiting scholar, I helped correct errors in the recorded original positions of some murals.

The collaboration was a mutual effort for resource sharing. The curator of the Museum of Asian Art first visited the Kizil Caves in 1998. Besides mutual visits, we shared the results of digital scans. They provided us with materials we lacked, and we launched projects for their references. In 2020, my book A Study on the Restoration of the Kizil Grotto Murals was published, thanks to the exchanges.

How is digital restoration conducted? What challenges have you faced?

I once felt I had completed my work after publishing the book. But I later found that many murals were heavily damaged. Some were destroyed during the long-distance transportation back to China. Moreover, due to change of religions (from Buddhism to Islam) in Xinjiang, the Buddhist monasteries in the Kizil Caves were left unmanaged and some of the Buddha figures collapsed.

This inspired us to pursue digital restoration. The original murals were resplendent, and we wanted to restore them to their former glory. We applied for a project with the regional authorities, focusing on restoring Cave 38. Most of the cave has now been restored.

Last year, we collaborated with China's tech giant Tencent, and they introduced the Tanyuan Plan, an initiative for cultural preservation through digital solutions. We used AI to restore the patterns in Cave 38, such as triangular hanging tent patterns and aquatic animals. The patterns are repetitive, which makes manual restoration tedious and time-consuming, whereas AI is excellent for this task.

However, AI restoration also presents challenges. Since the patterns are consistent, the AI model can restore the unclear parts drawing inference from the clear parts. However, this approach does not work for human figures, as each figure's expression is unique. Also, when we attempted to restore the scenes in Cave 38, the results were unsatisfactory, as the restored images looked too new.

Before manual restoration, we created a small dataset for training the AI model. In 2025, I applied for a project to create a dataset for rhombus-patterned murals, laying the groundwork for model recognition.

How do you balance tourism development and cultural relic protection?

While overdevelopment of tourist sites may compromise preservation, the very purpose of protecting cultural heritage is to make it accessible to the public. Our approach seeks to balance conservation with public access. Currently, only six of the caves are open to visitors, with each guided group restricted to no more than 10 people and a total daily visitor capacity of 1,340.

We are constructing a 9,800-square-meter digital exhibition center to showcase our digital restoration work, with the primary aim of safeguarding the physical relics while enhancing visitor understanding. By offering an immersive digital experience, it will help alleviate pressure on the actual caves, addressing frequent challenges of overcrowding and ensuring long-term preservation.

Since many visitors struggle to understand the murals, digital exhibitions enable them to see the original caves and help them comprehend. This would not prevent visitors from visiting the caves. Instead, after seeing the digital exhibition, they would be even more eager to visit the caves.

We are now working on digital restoration and planning exhibitions, including overseas tours. We are organizing an exhibition that we hope to premiere in Beijing next year and later tour internationally. We are also developing digital cultural products and websites for overseas audiences.

To attract international tourists and foster cooperation, the most important thing is to let them experience it firsthand. I organized an international conference last year, before which many foreign experts asked if Xinjiang was safe. I told them Xinjiang is now one of the safest regions in the world. After they came, they saw the real Xinjiang for themselves. 

Comments to [email protected]

SOURCE Xinjiang Today