Motion Warning. This video includes abrupt camera movement or other effects. Stop viewing immediately if you feel discomfort.
New: AI passthrough!
This amazing Deo feature uses the power of AI to turn every VR scene into AR passthrough! Now you can take characters out of VR and have them right there with you - as if they were in the same room.
Notice: AI Passthrough is presently in beta mode, and as such, users may encounter occasional service imperfections. The feature is currently exclusive to the DeoVR app, but it will soon be accessible on both browsers and mobile devices. Your feedback is highly encouraged and appreciated.
Recommended headsets:
Meta Quest 3, and Quest Pro with stereoscopic color passthrough, Pico 4 (monoscopic color passthrough).
Compatible headsets:
Quest 2, Valve Index (monoscopic black and white passthrough).
Passthrough is not compatible yet for Oculus Link cable.
Check out our complete guide to passthrough and join in the discussion at our busy forum.
Haiyue Kaijin embodies the classical Chinese "one pond, three mountains" motif, drawing inspiration from the mythical Penglai. While traditionally rooted in the Qin Dynasty’s Lanchi Palace, this site distinguishes itself through its unique double-layered circular platform.
Unlike the prevalent "Chongtan Gaoge"—the square, high-platform structures that peaked during the Han Dynasty to symbolize a connection to the heavens—Haiyue Kaijin adopts a circular form. This choice reflects a profound respect for natural geometry and the specific functional requirements of island architecture. This spatial logic resonates with historical precedents such as the city-platform clusters of Beihai’s Tuancheng and the mandala-inspired symmetry of the Summer Palace’s Zhijing Pavilion. Together, these structures illustrate a shift toward ordered, centripetal layouts centered within radiating water features.
Regarding the reconstruction of the Old Summer Palace’s "Haiyue Kaijin" scene: While every effort has been made to maintain historical fidelity, please note that current limitations in archival data and modeling techniques may result in minor discrepancies.
