37.5 Holography🔗⧉

Everything in this part so far has treated light as rays, bundles of straight lines carrying intensity and color, sorted by direction. That ray picture is enough to explain stereo, parallax, and even the light-field displays of Chapter 37.4, Lenticular displays. But light is a wave, and a wave has something a ray does not: a phase, meaning where it is in its oscillation cycle at each point in space. When a wave reaches your eye, it arrives with a particular shape, a wavefront: the surface of constant phase. Light diverging from a near point arrives as a tightly curved wavefront; light from a far point arrives nearly flat. And it is precisely that curvature your eye's lens reads when it accommodates, because the eye focuses by reshaping the curved wavefront to a point on the retina. A flat screen, even a clever light-field one, only approximates the wavefront with a finite fan of rays. Holography reproduces the wavefront itself. That is the whole idea, and it is why holography is the one display where getting the physics right makes accommodation automatic rather than an extra feature to bolt on.

The difficulty is that ordinary intensity detectors do not measure optical phase directly: detectors (film, CCD, your retina) measure intensity, the squared amplitude, and throw the phase away. The defining trick of holography is a way to record phase as intensity and then play it back as a wavefront, and it rests on interference.

37.5.1 Lippmann and Gabor: recording the wave🔗⧉

The intellectual root is again Chapter 31.2, Light field cameras, the same physicist behind integral photography. In 1891 Gabriel Lippmann invented interference color photography Lippmann 1891, which won him the 1908 Nobel Prize. He recorded color not with dyes but with standing waves: light reflecting off a mirror behind the emulsion interfered with the incoming light, laying down microscopic layers in the film at spacings set by the light's wavelength, so the developed plate reconstructed color by reflecting back exactly those wavelengths. It was the first photographic process to record information in the standing-wave structure of light rather than in intensity alone, the conceptual seed of capturing phase.

Holography proper was invented by Dennis Gabor in 1948 Gabor 1948, for which he received the 1971 Nobel Prize. (He was trying to improve electron microscopy, not to make 3-D pictures; the 3-D display application came later.) Gabor's insight: bring the light from the object together with a coherent reference beam at the recording plate, and let them interfere. Where the two waves arrive in phase they reinforce (bright); where they arrive out of phase they cancel (dark). The resulting pattern of fine fringes is an intensity pattern an ordinary plate can record, but because the fringe positions depend on the phase difference between object and reference at each point, that intensity pattern secretly encodes the object wave's phase as well as its amplitude. The phase difference appears in a recordable intensity. With object field $U_o$ and reference field $U_r$,

The cross term carries their relative phase. A single intensity sample does not uniquely determine a general complex field; the known reference and the holographic encoding geometry make separation and reconstruction possible.

On replay, illuminate the developed fringe pattern with the reference beam again, and the fringes act as a diffraction grating. They bend the reference light by exactly the angles that reconstruct the original object wave. The wavefront that leaves the hologram is (to good approximation) the very wavefront the object once sent. Within the reconstructed aperture, wavelength band, and viewing zone, it can reproduce the corresponding visual cues: look from the left and you see the object's left side; move your head and it shows full, continuous parallax; and your lens accommodates to the true depth of each virtual point, because the reconstructed wavefront has the correct curvature.

37.5.2 Off-axis holography and the space-bandwidth wall🔗⧉

Gabor's original "in-line" arrangement had a flaw: the reconstructed image, the reference beam, and a spurious "twin" image all came out along the same axis, muddling the view. Emmett Leith and Juris Upatnieks fixed this in 1962 Leith & Upatnieks 1962, armed with the newly invented laser, which gave the coherent light holography needs, by bringing the reference beam in at an angle (off-axis). The angular offset separates the reconstructed image from the leftover beams, so a clean virtual image floats free. Off-axis holography with laser light is what turned holography from Gabor's curiosity into the vivid 3-D holograms of museums and credit cards.

But dynamic, computer-driven, room-sized holographic displays run into a wall that no cleverness removes, only manages: the space-bandwidth product. The fringes that encode a wavefront are spaced on the order of the wavelength of light, at sub-micron scale. To diffract light over a wide angle (so the image can be seen from a broad range of viewpoints, with wide parallax) you need the finest fringes; to make a large image you need the pattern to cover a large area. The number of optical degrees of freedom scales roughly as display area times supported solid angle divided by wavelength squared, and for anything like a comfortable, life-sized, wide-angle holographic image that number runs into the billions to trillions, orders of magnitude beyond the addressable sample counts of conventional spatial light modulators. The central obstacle for holographic displays is not reconstructing a wavefront (we can) but building a modulator with enough fine elements to reconstruct a big, wide one, which we cannot yet do. For pixel pitch $p$ under normal reference incidence, a simple one-dimensional sampling bound is $|\sin\theta_{\max}|\lesssim\lambda/(2p)$. Smaller pixels support larger diffraction angles; increasing panel width at fixed pitch increases the aperture but not this angular bound. Real holographic displays therefore trade away field of view, image size, or eyebox to fit the budget. This is why holographic near-eye displays (Maimone et al. 2017, in Chapter 37.3, 3D displays with accommodation) are promising: an eyebox the size of a pupil needs far fewer fringes than a room-sized window.

37.5.3 Computational holography🔗⧉

The final move is to stop recording holograms optically and start computing them. Instead of a laser, a beam-splitter, and a physical object, take a synthetic 3-D scene and compute a hologram encoding the desired field (a computer-generated hologram, or CGH), then display an appropriate phase encoding on a programmable phase spatial light modulator (Optical modulators (spatial light modulators): DMD and LCD/LCoS), typically a phase-modulating LCoS panel that imposes a controllable phase delay at each of its pixels. Illuminate the SLM with coherent light and it diffracts the beam into the wavefront of the synthetic scene. This gives a holographic display with no physical original, reprogrammable every frame.

The hard part is the inverse problem. We can easily simulate forward how a given SLM phase pattern diffracts (it is a well-understood wave-propagation computation, such as a Fresnel or Fraunhofer transform). What we want is the inverse: find the phase pattern whose diffraction equals the target image. That inverse problem is hard: phase-only modulators can set phase but not amplitude, the propagation is a nonlinear function of the phase, and naive solutions are speckled and dim. The modern approach treats it as an optimization: model the wave propagation as a differentiable function of the SLM phase, then descend the gradient of the error between the simulated reconstruction and the target, designing the optical field by backpropagation, exactly the move of Chapter 33.7, End-to-end optimization, with the SLM phase playing the role of a "designed degree of freedom" as in Chapter 33.1, Wavefront coding. Peng, Choi, Padmanaban, and Wetzstein's neural holography (2020) Peng et al. 2020 did this with a learned, camera-in-the-loop model that corrects for the real optics' imperfections; Shi, Li, Chakravarthula, and Matusik's tensor holography (2021) Shi et al. 2021 trained a neural network to predict the CGH directly from an RGB-D image, fast enough for real-time photorealistic 3-D holography. These are the techniques that, if the space-bandwidth wall is ever scaled, would drive a true holographic display, the closest thing to Sutherland's ultimate display, the endpoint this whole part has been climbing toward.