While 3-D displays have been the stuff of fantasy for years, working, useful, spatial 3-D is a reality. Spatial 3-D is breaking down a critical bottleneckthe displaythat regulates the flow from today's huge multidimensional databases into your brain. In order to function, however, such displays require electro-optical components capable of transforming digital signals into visual information at rates of gigabytes per second.
Hollywood movies have depicted spatial 3-D for many years, usually as "holographic" images projected into mid-air. While these theatrical constructs function only in a world unconstrained by the laws of physics, a wide range of dynamic spatial 3-D technologies exist today, including dome-shaped volumetric displays, electro-holographic systems, and parallax-based multiperspective devices.1 Although they differ greatly in technology, each of these display types works by approximating the 3-D field of light emitted by physical objects.2
Spatial light modulators (SLMs) make these displays possible by turning bits of image data into directed photons of light. The information bandwidth of the modulator and the signal driving it determine the fidelity of the displayed spatial image.
Since real-world scenes contain staggering amounts of visual information, SLM bandwidth is key. For instance, an uncompressed representation of a 0.5-m cubic volume (0.125 m3) sampled at 0.5-mm resolution in 12-bit color would consume 45 GB/s of data at video rates. High-quality display holograms have fringe patterns on the order of 1500 line pairs per millimeter over the entire display. Parallax displays typically need to display scores of video frames simultaneously. While compression allows this data to be transmitted more efficiently, the final step of image modulation generally requires an uncompressed stream. 3-D displays stretch the capacity of today's SLM technologies up to, and beyond, their limits.
One way to approach this requirement is to spatially multiplex the output of a high frame-rate SLM. In our 3-D system, a standard desktop PC allows the user to view volume-filling 25-cm-diameter 3-D imagery that hovers inside of a clear dome (see figure). First, 3-D structure is obtained from off-the-shelf software such as a molecular visualization program. Then, we slice the 3-D data into 198 segments, each at a resolution of 768 X 768. Finally, we optically reconstruct the volume using a digital light processor (Texas Instruments; Dallas, TX) as a 2-D SLM running at over 5000 frames per second. A bank of three SLMs projects the sequence of 198 slices onto a screen that sweeps a 3-D volume at 30 Hz. That's where high space-bandwidth-product SLMs are required: our system has an optical bandwidth of over 1.3 GB/s.
Another approach involves tiling the output of a fast electrically addressable SLM onto an optically addressable SLM (QinetiQ; Farnborough, UK).
A third method to squeeze performance out of high-bandwidth SLMs is angular multiplexing, in which the image of an SLM is swept across a range of angles in space.3 This technique is used in various view-sequential displays in which the bandwidth requirements are just as high, ultimately requiring on the order of 100 2-D bitmaps to reconstruct a realistic 3-D scene.
Spatial 3-D displays are enabling advances in visualizationnotably in medical research, when surgeons need to quickly pinpoint the location of a harmful mass, or in military visualization, when an autonomous aircraft is guided to a moving target. Since all of these applications produce gigabytes of visual information at video rates, low-cost, high-bandwidth SLMs will probably continue to be required for commercially viable 3-D displays. oe
1. B. Blundell and Adam Schwarz, Volumetric three-dimensional display systems, Wiley-Interscience (2000).
2. G. Favalora and D. Lewis, "Spatial 3-D: The End of Flat-Screen Thinking," Actuality Systems, Inc. whitepaper (1 July 2003).
3. M.W. Halle, "Multiple Viewpoint Rendering for Three-Dimensional Displays," PhD Thesis, Massachusetts Institute of Technology, Cambridge, Mass. (1997).
Gregg Favalora is CTO and VP, product of Actuality Systems, Burlington, MA.