Toshiki Hijiri

A compression/decompression method for streaming based humanoid animation

In VRML, a modeling language for describing 3D objects on the internet, the specification to realize lifelike movement of a 3D character with a skeletal structure (such as a human) has been standardized as VRML Humanoid Animation Ver.1.0 (H-Anim Ver.1.0) in the H-Anim WG of the VRML Consortium. To extend this specification, we suggest a method that makes it possible to send/receive motion data in real time on a network with narrow bandwidth such as a telephone line. Moreover, by sending the motion data with streaming data from server to client, the time required before playback can be greatly reduced. This technology uses basic techniques that can be applied widely to webbased 3D applications, broadcasting contents etc.

A spatial hierarchical compression method for 3D streaming animation

When distributing 3D contents real-time over a network with a narrow bandwidth such as a telephone line, methods for streaming and data compression can be considered indispensable. In previous work, we made possible the real-time streaming of 3D animation data on a network with a narrow bandwidth such as a telephone line by partitioning motion data for humanoid characters (data obtained by motion capture, for example full frame data at 30 frames/sec) into packets and then carrying out compression by culling data along the time axis. However, as a 3D scene becomes more complex, the number of humanoid characters also increases. Accordingly, the transmission rate also increases, becoming greater than the available bandwidth and making real-time distribution impossible. In this paper, we concentrate on the problem of real-time distribution, describing a new data packet format which allows flexible scalability of the transmission rate, and a data compression method, SHCM, which maximizes the features of this format using a 3D scene structure. Because compression using a 3D scene structure aims to obtain the optimal overall compression rate by altering the compression rate for each object, based on information on the position in 3D space relative to the behavior (motion) data of each object, its application to MPEG4 can be expected. Using this method the real-time distribution of 3D contents becomes possible despite the bandwidth restrictions of an ordinary telephone line.