This paper discusses the implementation of live video transmission in MPEG-4 3D scene. A captured video stream is packetized and time-stamped in MPEG-4 synchronization layer (SL) and transmitted on the channel created by DMIF remote instance. The primary method is to parse the MP4 file and utilize the characteristics of elementary stream (ES) descriptor. The SL configuration and SL header are designed for the live video such that the live media can be properly transmitted and combined with other objects in the original 3D scene. The implemented system has been well verified with the integration version of IM1 3D player and UBC’s DMIF.