Streaming and Rendering of 3D Dynamic Scene in a Network Virtual Environment by Predictive Modelling

A 3D scene in the virtual environment is dynamic by varying either the scene position as a whole or a part of scene parameters (location of vertices, connectivity etc.) as a function of time. In the proposed work, the object position in a dynamic scene changes temporally and is visualized by static or dynamic camera positions. This contrasts with the static scene where the camera position alone is dynamic, and the object is static. The camera movements are controlled by the user’s interactions. In this study, the object’s temporal position is being considered along with the user navigation pattern and eye/camera positions for building the predictive engine. The real challenge lies in building the predictive engine based on the user interaction dynamics which changes with respect to the object position. Initially, dynamic scene streaming and rendering with predictive approach is attempted at object level and then further extended to scene level. Again, the focus is to improve the streaming and rendering performance with the available bandwidth and memory capacity. Since the objects in the scene are dynamic in nature and the Level of Detail varies at different positions, the hybrid predictive-progressive streaming and rendering approach has been adapted. The experiment results are encouraging and could achieve considerable reduction in mesh streaming and rendering thus saving enough bandwidth and memory.