Projection display using computer-generated phase screens

The first device based on a phase screen for projection display was the Eidophor system, initially developed in the 1940s.¹ An oil film is deformed by a cathode ray beam to produce the phase profile, while a so-called Schlieren system is used to convert the phase image into an amplitude image. Schlieren optics result in light loss because the light diffracted from the beam path is deposited on the bars of a mirror bar system. In a more efficient phase structure, light is redirected from the dark to the illuminated parts of the image. Today, more complex phase screens than the ones used in the Eidophor system perform this operation.

At a meeting with contractors in December 2000, two of us (NC and BC) raised the prospect of using phase-modulating screens in place of amplitude-modulating screens for 3D holographic display. The holograms written on the existing amplitude-modulating screens reconstructed a 3D scene in the far field with the help of a concave mirror. We suggested replacing these with phase-modulating screens to enhance diffraction efficiency. At the time, this proposal was not pursued because developing the phase hologram required an extra computational step. In addition, the replay field would have had a lower signal-to-noise ratio. Reduction in computational overhead was a major objective due to the high level of such resources required for 3D holograms.

Applying this concept to projection of 2D scenes, however, the same advantage accrues at a lower computational cost. In fact, as was known by the 1990s, specialist digital processing hardware allows the phase screen to be computed in real time.² It is a phase-only Fourier hologram of the scene computed using iterative techniques based on the Gerchberg–Saxton algorithm.³

Figure 1. (a) Fresnel zone plate, (b) phase zone plate, and (c) phase Fresnel lens.

A spatial light modulator is the key component for modulating the phase of the light beam. The device we use is a nematic liquid crystal on silicon (LCOS) integrated circuit, which was initially researched in the 1970s.⁴ Alternative devices include those based on piston microelectromechanical systems (MEMS), such as the Grating Light Valve.⁵ We currently use an HDTV LCOS device from Holoeye Photonics AG which produces 2pi phase modulation in the visible region. It is important that the device generate a full wave of phase modulation to maximize efficiency.

Figure 2. Projected image from an amplitude screen discretized to 4bit gray levels (left), and from a phase screen discretized to 4bit phase levels (right).

Figure 3. Replay field of hologram using R, G, and B lasers.

Figure 1 shows three approaches to making a screen to accomplish a simple task, such as focusing a collimated light beam. The Fresnel zone plate (a), or FZP, is a binary amplitude screen composed of absorbing and transmitting zones that produce a diffraction pattern along the optical axis. Approximately 10% of the incident light goes into the primary focus. The phase zone plate (b) is a binary phase screen with no absorbing zones. The alternating zones of 0 and pi phase delay can be viewed as the sum of the transmitting zones of two FZPs. Thus, the resulting amplitude efficiency in the primary focus is twice that of the FZP and the intensity efficiency is four times, or 40%. Finally, the phase Fresnel lens (c), an analog phase screen, is a modulo 2pi equivalent to a planoconvex lens that will produce up to 100% efficiency at the focus of the lens.

In addition to redistribution of light in the image plane, there is a further efficiency advantage due to the use of polarized solid-state light sources. R, G, and B sources provide a rich color gamut with no requirement for an absorbing color wheel or polarizer.

Moreover, due to the Fourier transform relationship between the hologram plane and the replay field, display brightness is improved by using a small display device. This is the opposite situation to conventional projection displays where the etendue consideration obliges manufacturers to maintain a larger area for the display device.⁶ Small device area reduces silicon real estate and the cost of volume production.

Smaller pixel size is also avoided in conventional displays because it entails proportionally larger pixel gaps that reduce the cosmetic quality of the display. In phase projection, the gaps between the pixels in the replay field are controlled by the degree of tiling in the hologram plane. Small device pixels increase the hologram resolution and strengthen fringing fields, which may improve overall phase response.

The effect of quantization in the hologram plane is shown in Figure 2. Quantization noise is redistributed throughout the replay field so that it is no longer visible compared with the amplitude quantized image. Similar considerations apply in the case of `dead’ pixels due to faults in the display device, so that yield margins improve when used for phase projection. Further benefits are frame-by-frame averaging of the noise field and software control of replay field aspect ratio.

We have constructed an optical testbed for an LCOS device in which the phase image reflected from the device is propagated through a lens onto a screen (see Figure 3). The replay field shows good registration of the three colors despite the fact that a single projection lens is used. This attests to the versatility of holograms that can be individually tuned to generate a color display.