An Adaptive Low-cost Approach to Display Different Scenes to Multi-users for the Light Field Display

As a promising alternative to VR head-mounted displays, current autostereoscopic displays and light field displays require high GPU consumption for multiple-view rendering and do not display different scenes for multiple viewers with motion parallax. Building upon prior work demonstrating how GPU utilization can be reduced by only rendering visible views, we propose an innovative approach to only render the visible views of different scenes towards multiple viewers’ eyes. Moreover, we found that the number of visible views decreases as the viewing distances increase. Thus, a dynamic approach can be taken to adjust the number of rendered views according to viewers’ distance from the display. This approach can be easily adapted to the off-the-shelf light field displays to display different 3D scenes for at least two viewers according to their head positions with reduced GPU costs.

This work is a prototype published at VRST 2020. The publication of this work could be found here.