The Problem

A computer-generated hologram (CGH) is a digitally generated pattern which can be reconstructed via holographic interference to form a 2D or 3D image.

A CGH can be printed on a photographic mask or film, or displayed on a reflective or transmissive microdisplay (also known as a Spatial Light Modulator or SLM).

For this to work well, the feature (or pixel) size has to be comparable to the wavelength of the light which will be used for its reconstruction – typically, about 500 nm. In round numbers, something like 50,000 dpi is ideal, but that’s 100 times the resolution of a high-end cell phone.

This kind of resolution is actually quite easy to achieve using analog photographic means, but doing so with a real-time digital microdisplay is very challenging. Not only must the pixels be tiny, but the computational bandwidth required to calculate and drive them is immense… CGH patterns can’t easily be approximated, and they don’t compress well.

However, traditional photographically-recorded holograms are limited because the object they record has to actually exist, be accessible, interact with laser light, and not move more than about a millionth of an inch during the recording process. Doing this well requires complex, expensive, delicate laboratory equipment. In contrast, the CGH approach can be used to show any kind of image, on-the-fly.

The Potential

Moving color 3D holograms are the most realistic possible images for immersive AR and MR. They can appear at any distance from the user, and are visually indistinguishable from reality. As they move closer to the user, they exhibit correct “vergence” and “accommodation”, the key visual cues which directly inform the human visual system how close something is.

Holographic light manipulation can also be used for many non-display applications, such as optical tweezers for trapping and manipulating biological samples, and wavefront correctors to allow ground-based telescopes to exceed the resolution limits imposed by the atmosphere. They can even be used to replace bulky and expensive optical elements like lenses and mirrors, in optical systems such as projectors, automotive Heads-up-Displays (HUDs), and VR/AV/MR headsets.

The State-of-the-Art

Holographic (or phase) microdisplays are now available from several different manufacturers with small-enough, good-enough pixels – and enough of them in one microdisplay. These devices are starting to be designed into next-generation headsets and HUDs. The electronic and economic challenges are certainly non-trivial, but commercially viable solutions do now exist.

And plenty of potential users have suitable “content” for such a display, from surgeons planning delicate procedures guided by 3D CT and MR scans, to gamers wanting the best possible immersive experience, and drivers needing 3D HUDs so that they can safely concentrate on the road. Clarify provides “holographic middleware” which bridges the gap between display hardware and user content. Our algorithms compute efficient CGH patterns for 2D and full 3D display, including full-color, and can be accelerated for real-time rendering using DSPs, GPUs, FPGAs, and ASICs.

Traditional CGH Computation and Challenges

Broadly speaking, there have been three main methods of CGH generating:

  • Point source holograms This computational strategy is based on analogies with the Huygenian Principle and point-source optical concepts. The object is broken down into a huge number of self-luminous points (a “point cloud” or “light cloud”). An elemental CGH is calculated for every such point, and the final composite CGH is synthesized by mathematically superimposing all these elemental holograms. For some subjects this is a satisfactory approach, but there’s no “occlusion” in surfaces made from points this way – every point can be seen all the time, even when it should be hidden behind another surface.

  • Fourier transforms Every point in a true hologram image is formed by the interaction of light deriving from every point on the CGH. Mathematical functions can be used to transform between a 2D representation (the CGH on the microdisplay) and the 3D image seen by the user. The Fourier approach produces sharp images quickly, but only at infinity… the image can’t be brought up close, and hence depth cannot be seen.

  • Fresnel transforms The Fresnel approach can produce a near-eye image, showing objects floating in space, but it’s always been much slower, and the approximations used have degraded the image.

Clarify’s Improved Solution

Clarify’s advanced algorithms uses a new approach to generate a Fresnel CGH with very high accuracy and low noise, at very high speeds.

For more details, please see our online CGH PowerPoint

Or get started immediately – we invite you to generate a CGH from your own 2D or 3D content, using free-to-try version of our algorithms (login required). Our web front-end allows you to upload a 2D image or a 3D CAD model, select an algorithm type and options, select the resolution and other properties of your SLM, and then download a free CGH.

Of course, you’ll need an SLM and suitable optics to reconstruct your CGH — by eye, it will just look like white-noise.

And some of our advanced features do require an account :) Please contact us for these advanced features, and for more technical information, consulting services, and licensing details.