We describe a device which has the potential to be used both as a virtual image display and as a backlight. The pupil of the emitted light fills the device approximately to its periphery and the collimated emission can be scanned both horizontally and vertically in the manner needed to illuminate an eye in any position. The aim is to reduce the power needed to illuminate a liquid crystal panel but also to enable a smooth transition from 3D to a virtual image as the user nears the screen.
©2013 Optical Society of America
The display of three dimensional images without the need for spectacles has been greatly simplified by the long anticipated arrival of devices such as Microsoft’s Kinect which reliably track the heads of those who watch. No longer must a display blindly show views at discrete angles to all potential viewing positions  but instead one view can be made visible to each eye at a time . Similarly, holographic video need no longer be synthesized from many narrow-angle constituents  and one low resolution hologram per eye will in principle suffice .
If the viewer is not to wear special spectacles, some way must be provided of making each view on the liquid crystal display visible to no more than one eye at a time. Replacing the backlight of a liquid crystal display with a display-sized lens and a spot source of light near the focal plane of the lens is a simple approach but it is bulky. Several attempts have been made to produce slim backlights which produce collimated light in such a way that it can be scanned. An appropriately shaped wedge cleanly collimates light from any light emitting diode along one edge but since the edge is one dimensional, light can be scanned either side to side or up and down but not both . An array of fluid prisms can scan light using electro-wetting but this approach is an ambitious departure from conventional backlight technology .
A display in the focal plane of a lens is often used to create a virtual image, i.e. an image which appears much more distant than its source. Such displays appear in cockpits reflected off the canopy so pilots can keep their heads up while reading their instruments but virtual image displays are also found in the eyepieces of video cameras, for example. Just as with backlights, designers of virtual image displays want their devices to be slim and an approach which is gaining attention is to direct a ray into a slab light-guide embossed with a weak grating, i.e. a grating which diffracts only a small fraction of incident light [7–10].
Each time a guided ray reflects off a surface embossed with a grating of appropriate pitch, part of the ray is diffracted out of the light-guide as shown in Fig. 1. If the pitch of the grating and the thickness of the guide are constant, the diffracted components will all emerge in parallel. To an eye looking through the grating, the components will appear to come from an infinitely distant point and therefore constitute a single virtual pixel. Alter the angle between the ray and the normal of the guide and the angle of the diffracted components will alter, so it is possible to scan a line of virtual pixels.
The temptation for the designer of a virtual display is simply to point a video projector into such a light-guide but rays within the guide will fan out from the projector so that nowhere is the entire virtual image visible. The challenge is how to manage this problem. One approach is to emboss the light-guide with a preliminary grating which recycles rays back to the center of the light-guide, but this means that the entire virtual image is available only towards the center of the guide.
Proposed here is a solution which draws on mechanically scanned projectors where a laser beam is scanned first by a mirror oscillating about, say, a vertical axis then by a mirror oscillating about a horizontal axis so as to trace a Lissajou pattern in the far field. Provided that the mirrors oscillate sufficiently fast, the laser beam can be modulated so as to project an image onto a distant screen. The pupil of such a projector is no bigger than the diameter of the laser beam so we expand it first by passing it into a light guide shaped like a rod (i.e. long and thin as in the left hand of Fig. 1) which is embossed with a grating so as to expand the pupil in the direction of travel of the beam. Then we reflect the diffracted components off the second scanning mirror and couple them into a slab light-guide embossed with a grating so as to expand the pupil in the orthogonal dimension (right hand of Fig. 1, and Fig. 2). A consequence is that the second scanning mirror must span the width of the slab so is much bigger than in a scanned beam projector. But although big, the second mirror scans about its long axis so its moment of inertia increases linearly with length rather than with the square or cube of any dimension. It is therefore not difficult to scan it at frequencies significantly greater than the ~70 Hz threshold where flicker is usually perceived.
The advantage of this approach over previous light guide concepts is that instead of placing the pupil expansion optics after the projector, we have interleaved one with the other. By initially scanning the ray in only one dimension, we can expand the pupil in that dimension without any fan-out in the orthogonal dimension. When rays emerge normal to the slab, they therefore do so from its entire surface. Most rays of course travel within the slab at an angle to its central axis, leaving one or other corner of the slab in shadow. This can in principle be corrected by injecting also rays at the opposite angle so that they reflect off the side of the slab and fill the shadowed corner, as explained in .
Borofloat glass was cut and polished into a rod (40 x 3 x 1 mm3) and slab (40 x 40 x 1 mm3) with a 30° bevel at one end of the rod and a 49° bevel for the slab. Using UV-curing glue, each was centrally embossed with a 25 mm long grating of pitch 2400 lines/mm (a pitch so great that only the first order is generated). The unmodulated red beam (wavelength 650 mm) from a laser diode was shone in the flat end of the rod and reflected off a front-silvered mirror adjacent to the bevelled end. Components diffracted out of the rod were allowed to reflect off a 30 mm long front silvered mirror back through the rod and into the bevelled entrance of the slab. Small motors were attached to each of the mirrors and they were driven with a sinusoidal waveform so as to rotate clockwise then anti-clockwise as depicted in Figs. 1 and 2 at slightly different frequencies of approximately 70 Hz.
Figure 3 shows a photograph of the pattern created. The ellipse comprises light which has been scanned by both mirrors and its field of view was approximately 25° in the horizontal and the vertical. The horizontal line comprises light which has been diffracted downwards by the rod grating in Fig. 1(a) which therefore enters the slab without having been scanned by the long mirror. The shadow which clips the top of the ellipse is a result of the long mirror reflecting some rays into the slab at greater than the critical angle. In a correctly aligned system, both rod and long mirror would be slightly rotated in opposite directions about their long axis so that scanned light is completely guided by the slab (such that the ellipse is no longer clipped) while un-scanned light is beyond the slab’s critical angle (such that the horizontal line disappears). The faint duplicate ellipses in Fig. 3 are caused by light in the rod reflecting off one side of the guide before reaching the small scanning mirror shown in Fig. 1(a).
Figure 4 shows a photograph of the panel when a diffuser was placed against its surface. The non-uniformities are due firstly to the periodic reflection of the guided rays, and secondly due to the decay of the rays as they propagate (from right to left in the rod and bottom to top in the slab). In similar work [7, 8] the first kind of non-uniformity is typically corrected by injecting light via a grating so as to fill the pupil of the guide, and the second is typically corrected by increasing the depth of the grating in the direction of ray travel. Similarly, techniques have been described and demonstrated [8, 9] for the management of green and blue light.
The emission from the guide tested here is collimated so a Fresnel lens or its equivalent must be added in order to concentrate light into the eye. However, there is then the classic problem that the viewer might move closer to or further from the backlight and the focal length of a lens is fixed. Instead Fig. 5 shows how at any instant, rays from two points on the screen will appear visible one to each eye but the rest of the screen will appear dark. Now begin to scan the ray and the visible points will move, eventually so that a point which was visible to one eye is now visible to the other. With appropriate timing, the liquid crystal will have had long enough to change state and by scanning the entire display in this manner, a different view, holographic or otherwise, may be made visible to each eye whatever its distance.
Although not needed for 3D, we would nevertheless like to concentrate rays into the user’s eye so as to save power and keep the cost of the lasers low. We propose therefore that the embossed grating be divided into areas which can be switched on and off. Figure 6 shows how we might scan the angle of the guided ray while, at any instant, we switch off the gratings everywhere except where rays are required to emerge. This should allow a convergent fan of rays to be multiplexed over time and the distance to the point where they converge could be altered simply by adjusting the timing of the process.
Switchable gratings have been configured using liquid crystal materials but few switch at more than a few kilohertz and the anisotropy of the liquid crystal makes it likely that some light will diffract out even when the grating is supposedly off. A more promising approach may be to switch off the gratings by using electric field or piezoelectric transducers to pull them back the 2 or 3 micron distance needed to remove them from the evanescent field. This approach has already been demonstrated by several authors to make displays [11–14] but a disadvantage is that the cost per lumen of lasers is approximately one hundred times that for LEDs. Our hope is that by concentrating rays of light into the eyes of all users, we can reduce the optical power budget to near zero and therefore make the use of lasers affordable.
We have proposed and performed a simple test of a device which can both display a virtual image and act as the backlight for a liquid crystal display. The device uses a simple pair of mirrors to scan the emission in the vertical as well as in the horizontal, thus allowing a more efficient concentration of illumination into the eyes of each viewer than a device which scans only in the horizontal. The device is powered by lasers, a technology normally considered too expensive for use in backlights, but the lasers need deliver so little power that we think they will be no more expensive than the light emitting diodes otherwise needed. We explain how switchable grating arrays developed elsewhere for 2D display might instead be used in a backlight to permit concentration of light into the users’ eyes, whatever their distance from the screen. Even without this stratagem, it should be possible to display 3D and holographic images without the user needing to wear spectacles. Because the backlight can also display a virtual image, we should be able to display such images even as the user brings their nose close to the screen, thus making the experience of a display more like looking through a window.
The authors would like to thank Consultantnet Ltd., 94 High Street, Linton, Cambridge CB21 4JT, UK for having put together the apparatus for the experiments reported here.
References and links
2. D. Ezra, G. J. Woodgate, B. A. Omar, N. S. Holliman, J. Harrold, and L. S. Shapiro, “New autostereoscopic display system,” Sharp Tech. J. 62, 10–14 (1995).
3. A. R. L. Travis, “The display of three-dimensional video images,” Proc. IEEE 85(11), 1817–1832 (1997). [CrossRef]
4. A. Schwerdtner, N. Leister, and R. Häussler, “A new approach to electro-holography for TV and projection displays,” SID Symposium Digest of Technical Papers 38(1), 1224–1227 (2007). [CrossRef]
6. A. Schwerdtner, N. Leister, R. Häussler, S. Reichelt, G. Fütterer, and A. Schwerdtner, “Eye-tracking solutions for real-time holographic 3-D display,” SID Symposium Digest of Technical Papers 39(1), 345–347 (2008). [CrossRef]
7. Y. Amitai, “A two-dimensional aperture expander for ultra-compact, high-performance head-worn displays,” SID Symposium Digest of Technical Papers 36(1), 360–363 (2005). [CrossRef]
8. T. Levola, “Novel diffractive optical components for near to eye displays,” SID Symposium Digest of Technical Papers 37(1), 64–67 (2006). [CrossRef]
9. H. Mukawa, K. Akutsu, I. Matsumura, S. Nakano, T. Yoshida, M. Kuwahara, K. Aiki, and M. Ogawa, “A Full Color Eyewear Display using Holographic Planar Waveguides,” SID Symposium Digest of Technical Papers 39(1), 89–92 (2008). [CrossRef]
10. A. Cameron, “Optical Waveguide Technology & Its Application In Head Mounted Displays,” Proc. SPIE 8383A, paper 13 (2012).
11. X. Zhou and E. Gulari, “Waveguide panel display using electromechanical spatial modulators,” SID Symposium Digest of Technical Papers 29(1), 1022–1025 (1998). [CrossRef]
12. T. Oguchi, M. Hayase, and T. Hatsuzawa, “Micromachined Display Device Using Sheet Waveguide and Multicantilevers Driven by Electrostatic Force,” IEEE Trans. Ind. Electron. 52(4), 984–991 (2005). [CrossRef]
13. R. R. Buczynski, S. Kufner, R. Kasztelanic, A. Sagan, I. Frese, T. Klotzbucher, and T. Szoplik, “Polymer-based pixel matrix display in MOEMS technology,” Proc. SPIE 5455, 36–43 (2004). [CrossRef]
14. Y. Takeuchi, T. Nanataki, and I. Ohwada, “Tiled display utilizing ceramic actuators for large area flat panels,” SID Symposium Digest of Technical Papers 32(1), 470–473 (2001). [CrossRef]