Data from static mural drawings and motion capture of professional dancers were used to train a machine-learning model, which generated a human-computer collaborative choreography of animated movement sequences. Through this process, Cave Dance breathes new life into static dance paintings and translates ancient texts; it presents a contemporary reconstruction—and imagination—of these ancient dances.
Flowering from the diverse body of research of this interdisciplinary project, Cave Dance manifests as a set of digital installations that elucidate the multifold culture of dance in Buddhist cave shrines. The exhibition not only immerses the audiences in a dynamic world of Dunhuang dance, but also leads them into the deeper cultural dimension of Buddhist dance—where audiences are invited to contemplate the themes of body, life, and spiritual transcendence embodied by the celestial dance in the cave.

Born from the largest project team at CAMLab, Cave Dance brings together art historians, A.I. scientists, data scientists, musicologists, choreographers, Buddhist scholars, architects, and digital artists in a joint effort to bring the ancient Buddhist dance culture to life.