Hologram design is the process of engineering a surface or digital pattern that can reconstruct a three-dimensional light field, and the field has changed dramatically since Dennis Gabor first demonstrated the principle in the late 1940s. What was once a purely optical exercise involving lasers and photographic plates now spans computer-generated algorithms, neural networks, metasurface engineering, and even acoustic waves. The core challenge remains the same across all of these approaches: encoding enough information about an object’s light into a recording medium so that a viewer perceives realistic depth, color, and parallax.
How a Hologram Records Light
Every hologram starts with an interference pattern. When two beams of coherent light overlap inside a recording medium, they create alternating bands of high and low intensity. The spacing and angle of those bands depend on the directions the two beams travel, and the pattern they form is essentially a fingerprint of the light that bounced off the original object. If that pattern gets permanently stored in a material, it becomes a diffractive structure that can redirect a later beam of light to recreate the original wavefront.
The physical mechanism that locks the pattern in place varies. In traditional silver-halide emulsions, light-sensitive particles darken in the bright regions, creating an amplitude grating. In photorefractive crystals and certain polymers, the bright regions change the material’s refractive index instead, producing a phase grating that redirects light without absorbing it. Phase gratings tend to be more efficient because they waste less light energy on absorption. In either case, the result is a volume Bragg grating: a three-dimensional lattice of refractive-index or absorption variations that selectively diffracts light at the correct angle to reconstruct the recorded scene.1RP Photonics Encyclopedia. Holography
Gabor’s original holograms predated the laser, and the lack of a good coherent source stalled the field for more than a decade. Once lasers became available, the idea “took off to new and unsuspected applications,” as one retrospective put it, and the basic optical recording method remained the gold standard for decades.2Optica Publishing Group (Applied Optics). Gabor’s holography: seven decades influencing optics
Computer-Generated Holograms
Designing a hologram no longer requires an actual object sitting in a laser beam. Computer-generated holography (CGH) calculates the interference pattern mathematically and then displays it on a device or prints it onto a surface. The designer specifies a target image or 3D scene, and an algorithm works backward to figure out what pattern of phase and amplitude values a hologram plane would need in order to diffract light into that scene.
The most widely used family of algorithms descends from the Gerchberg-Saxton (GS) approach. These iterative methods simulate light bouncing back and forth between the hologram plane and the image plane, applying known constraints at each pass. After enough iterations, the algorithm converges on a phase pattern that produces a reconstruction close to the target image.3PubMed. Iterative phase retrieval for digital holography: tutorial In practice, though, “perfect” phase-only holograms for arbitrary target images remain elusive. Every phase-retrieval algorithm ends up with some residual error between what the designer wanted and what the hologram actually reconstructs, except for very specific target patterns.4University of Cambridge Research Outputs. Information capacity of phase-only computer-generated holograms for holographic displays
That gap between the ideal and the achievable is where much of the creative and engineering challenge of hologram design lives. Designers make trade-offs: sacrificing uniformity in brightness for sharper edges, accepting slight color shifts to keep computation time manageable, or limiting the viewing angle to boost image quality within a narrower cone.
Speeding Things Up with Neural Networks
Classical iterative algorithms are accurate but slow, especially at high resolutions. A single 4K hologram can require dozens of GS iterations, each involving a computationally heavy Fourier transform. That makes real-time holographic video impractical on consumer hardware using traditional methods alone.
Deep learning has changed the equation. One approach demonstrated a convolutional neural network small enough to fit in about 620 kilobytes of memory that could generate full-color 3D holograms at 1,920 × 1,080 resolution at 60 frames per second on a single consumer-grade GPU. The network was trained using loss functions that physically model how light diffracts, so the output patterns obey the same wave optics the iterative algorithms do, just much faster.5Nature. Towards real-time photorealistic 3D holography with deep neural networks
More recent work has pushed this further to 4K resolution. A network with only 938 parameters can generate a 2D 4K hologram in about 18 milliseconds, with each additional depth layer adding roughly 12 milliseconds. The reconstructed images scored above 30 dB on peak signal-to-noise ratio across a working depth range, which is solid image quality by display standards.6Scientific Reports. Real-time 4K computer-generated hologram based on encoding conventional neural network with learned layered phase
The trick these networks use is learning a set of “initial phase” templates during training that serve as universal starting points for any target image at a given depth. Instead of iterating from scratch every frame, the network essentially fine-tunes a pre-learned solution, which is why it can be so small and fast. For hologram designers, this means the computational bottleneck that once required dedicated cluster hardware is increasingly becoming a single-GPU problem.
Another Way to Speed Computation
Neural networks are not the only shortcut. When a hologram needs to represent a real physical scene rather than a synthetic one, designers can use depth cameras to capture the scene as a point cloud, then bin those points into layers at discrete depths. Because each layer can be processed with a fast Fourier transform instead of calculating every point individually, the computational load drops dramatically compared to brute-force point-source methods.7Optics Communications. Fast calculation method of computer-generated hologram using a depth camera with point cloud gridding This gridding technique and neural-network approaches are not mutually exclusive; some pipelines combine depth-camera capture with learned encoding to get both speed and quality.
The Speckle Problem
Anyone who has looked at a laser pointer dot on a wall has seen speckle: that grainy, shimmering texture caused by coherent light waves interfering randomly after scattering off a rough surface. In holographic displays and microscopy, speckle is the single most persistent image-quality nuisance. It obscures fine detail, reduces contrast, and makes text difficult to read.
Designers tackle speckle from multiple angles. On the computational side, adaptive 3D filtering methods process a stack of holograms, each carrying an independent speckle pattern, and average them intelligently. One recent technique using an adapted 3D Frost filter reduced the normalized standard deviation of speckle by up to 40% and improved structural similarity by up to 60% compared to conventional 2D and 3D filtering methods.8PubMed Central. Speckle Noise Reduction in Digital Holography by 3D Adaptive Filtering
On the hardware side, a framework called HoloChrome attacks speckle at its source by using many discrete wavelengths of light rather than just three. An ultrafast, wavelength-adjustable laser paired with a dual spatial-light-modulator setup multiplexes dozens of wavelengths across the visible spectrum. Because each wavelength produces a statistically independent speckle pattern, averaging them incoherently smooths the grain out of the final image.9ACM Transactions on Graphics. HoloChrome: Polychromatic Illumination for Speckle Reduction in Holographic Near-Eye Displays The trade-off is system complexity: more wavelengths mean more hardware and tighter calibration, but the perceptual payoff in image cleanliness is substantial.
Choosing a Recording Material
For holograms that need to exist as physical objects rather than live on a screen, the recording material matters enormously. Silver-halide films offer high resolution but require wet chemical processing. Dichromated gelatin gives excellent clarity but is fragile and humidity-sensitive. Photopolymers have emerged as a popular alternative because they can be self-developing, meaning the hologram forms during exposure without a separate chemical step.
One area of active development is making photopolymers from more sustainable and robust base materials. Researchers have demonstrated a cellulose acetate-based photopolymer that can record volume phase holograms at spatial frequencies of 800 line pairs per millimeter. Crucially, layers of this material survived being immersed in water for an hour with no measurable change in diffraction efficiency or surface quality.10Photonics. Water Resistant Cellulose Acetate Based Photopolymer for Recording of Volume Phase Holograms That water resistance opens the door to applications where holograms might be exposed to moisture, from outdoor security labels to biomedical sensors.
Photorefractive polymers occupy a different niche: they are rewritable. A large-area photorefractive polymer has been used to demonstrate holographic telepresence, refreshing holographic images every two seconds using a pulsed laser. Multicolored, full-parallax 3D images were achieved by combining angular and spatial multiplexing.11Nature. Holographic three-dimensional telepresence using large-area photorefractive polymer Two-second refresh rates are nowhere near video speed, but the demonstration showed that updatable physical holograms are possible, not just theoretical.
Full-Color Design with Metasurfaces
Achieving faithful full-color holography is harder than it sounds. A standard hologram designed for red light will not reconstruct correctly under green or blue illumination because the diffraction angles change with wavelength. Multi-color holograms traditionally require separate recordings for each color channel, which means either stacking multiple layers or accepting crosstalk between channels.
Metasurface holograms offer an elegant workaround. These are flat surfaces patterned with sub-wavelength structures, typically nanopillars made from a high-refractive-index dielectric. By carefully shaping each nanopillar, designers can independently control the phase imposed on red, green, and blue light passing through a single element. A recent demonstration used cross-shaped silicon nitride nanopillars to achieve wavelength- and angle-multiplexed full-color 3D holography from a single layer of polarization-independent meta-atoms, surpassing the pixel-density limits of earlier wavelength-multiplexing schemes.12PubMed Central. Wavelength- and angle-multiplexed full-color 3D metasurface hologram
Metasurface holograms are still largely a research-lab product. Fabricating sub-wavelength pillar arrays across large areas requires electron-beam lithography or advanced nanoimprint techniques, which are slow and expensive. But they represent a direction where hologram design converges with nanophotonic engineering, and the potential for extremely thin, lightweight, mass-producible holograms makes them attractive for future consumer products.
Holographic Near-Eye Displays
Augmented-reality glasses are arguably the application driving the most investment in hologram design today. Compared to other 3D display methods, holographic displays have a unique advantage: they can reproduce natural depth cues and even correct for a viewer’s optical aberrations, which reduces eye strain during prolonged use.13Optica. Toward the next-generation VR/AR optics: a review of holographic near-eye displays from a human-centric perspective
A major engineering hurdle for AR glasses is the eyebox, the region in space where your eye can sit and still see the full image. Traditional holographic displays have a tiny eyebox because diffraction angles are limited by the pixel pitch of the spatial light modulator. One solution feeds the hologram through an exit-pupil-expanding waveguide, the same kind of thin glass lightguide already used in commercial AR headsets. Researchers have shown that waveguide-coupled holography can provide a large, software-steerable eyebox while also suppressing phase discontinuities caused by the pupil-replication process, which improves resolution.14Nature Communications. Waveguide holography for 3D augmented reality glasses
Eye safety is another design constraint unique to near-eye holograms. Because a laser source sits millimeters from the retina, designers have to ensure that no failure mode could deliver a dangerous dose of light. Analysis has shown that holographic projector efficiencies can range from 50% down to 0.05%, averaging around 3%. An example projector built with just 1.78 milliwatts of laser power achieved a mean luminance of 30,000 candelas per square meter, and peak luminance above 300,000 candelas per square meter for sparse content. Those brightness levels are more than enough for daylight-readable AR, and the low input power keeps the system intrinsically safe for the eye.15Journal of the Society for Information Display. An intrinsically eye safe approach to high apparent brightness augmented reality displays using computer‐generated holography
Spatial Light Modulators as the Display Engine
Most real-time holographic displays rely on a spatial light modulator (SLM) to present the computed hologram pattern. The dominant technology is liquid crystal on silicon (LCOS), which works like a tiny reflective LCD: each pixel independently shifts the phase of reflected light by controlling the orientation of liquid-crystal molecules. Frame rate, phase linearity, and flicker are the key performance metrics that constrain what a hologram designer can achieve.16Photonics Letters of Poland. LCOS Spatial Light Modulator for Digital Holography
Pixel pitch is perhaps the most limiting factor. Current consumer LCOS panels have pixels on the order of a few micrometers, which caps the maximum diffraction angle and therefore the field of view. Designers compensate by using multiple SLMs, eye-tracking to steer the image, or the waveguide expansion techniques described above. Shrinking pixel pitch further is an active area of semiconductor development, and each halving of pitch roughly doubles the achievable viewing angle.
Holographic Microscopy
Hologram design principles feed directly into scientific instrumentation. Digital holographic microscopy (DHM) records the interference pattern between light passing through a transparent sample and a reference beam, then computationally reconstructs the sample’s phase map. Because biological cells are mostly transparent and barely alter the amplitude of light, conventional microscopy struggles to image them without staining. DHM captures the phase shifts that cells impose on transmitted light, revealing structure and thickness without any chemical labels.
When tested against a known calibration target, DHM measured thicknesses within about a tenth of a micrometer of the expected value, closely matching a competing phase-imaging technique measured simultaneously. The characteristic noise signature of DHM is grainy and speckle-like, distinct from the low-frequency artifacts of alternative methods, which means the two can complement each other when combined.17Optics and Lasers in Engineering. Quantitative phase imaging comparison of digital holographic microscopy and transport of intensity equation phase through simultaneous measurements of live cells For designers of holographic microscopes, this means that speckle reduction and noise characterization are not just display concerns but measurement-accuracy concerns.
Holography Beyond Light
The principle of recording and reconstructing a wavefront is not limited to electromagnetic waves. Acoustic holography applies the same interference logic to sound fields, and a recent extension called acoustofluidic holography combines designed acoustic patterns with controlled fluid flow to manipulate particles ranging from hundreds of micrometers down to tens of nanometers. The technique works on cells, polymers, and metal particles alike.18PubMed Central. Acoustofluidic Holography for Micro- to Nanoscale Particle Manipulation
The design process for an acoustic hologram mirrors optical CGH in many ways: the designer specifies a target pressure distribution and computes the surface pattern needed to produce it. The difference is that the “pixels” are physical surface features on a 3D-printed plate rather than electronically addressed modulators, and the operating frequencies are in the ultrasound range rather than the optical range. For biomedical applications like cell patterning and tissue engineering, acoustic holography has the advantage of being gentle enough to handle living cells without damaging them, something optical trapping struggles with at larger scales.
Where the Field Is Headed
The convergence of several threads is worth watching. Neural-network hologram generators are getting small and fast enough to run on mobile processors, which is exactly what AR glasses need. Metasurface fabrication is moving from electron-beam lithography toward nanoimprint methods that could eventually support mass production. Photopolymer materials are becoming more environmentally robust and chemically sustainable. And speckle reduction, long the Achilles’ heel of coherent-light displays, is being attacked from both the algorithm and hardware sides simultaneously.
One underappreciated aspect of hologram design is that the field sits at an unusual intersection of optics, materials science, computer science, and perceptual psychology. A hologram that is optically perfect can still look wrong to a human viewer if the depth cues conflict with what the visual system expects, or if color rendering shifts under different viewing conditions. Bridging that gap between physical accuracy and perceptual realism is where much of the artistry in hologram design still lives, and it is a problem that no single discipline can solve on its own.