Compact Lens Design Will Improve Cameras for AR-VR Devices
As we have said in earlier articles, the quality of the graphics displayed in XR devices is hugely dependent upon the camera-lens systems that are installed. In addition, these essential assemblies have an enormous impact on the overall size and dimensions of a headset, and this is important as hardware developers try and wring as much out of the smallest packages that they can achieve.
Because of this, manufacturers are always looking for ways to deliver the best possible performance in the smallest-possible package and if the rumours about the next Meta device being similar to sunglasses are true, then the assemblies need to be very small indeed.
Without doubt, companies like Samsung and Meta will be watching developments at the Seoul National University College of Engineering with interest. Researchers there are in the midst of developing a new camera that operates by using metasurfaces to deliver truly astounding content.
When combined with a folding lens system that makes use of these metasurfaces, this new optical design method is anticipated to result in a considerable reduction in the bulk of cameras. This has the potential to significantly reduce the optical side of XR headsets and allow systems to become smaller yet more powerful.
The Problem with Cameras
Conventional cameras are engineered to assemble many glass lenses to refract light during picture capture. This structure yields superior high-quality images; however, the lens thickness and considerable spacing between lenses add to the camera’s bulk, complicating its integration into ultra-compact devices like virtual and augmented reality apparatuses, as well as related devices such as smartphones. Tiny cameras mean smaller products.
Current image systems, including those utilised in smartphones, virtual reality, and augmented reality gadgets, are continuously advancing to achieve greater compactness, efficiency, and performance. Conventional optical systems depend on substantial glass lenses, which exhibit limitations such as chromatic aberrations, reduced efficiency across various wavelengths, and considerable physical dimensions.
But the answer doesn’t lie in the miniaturisation of camera devices as they stand. There is only so much that even the most skilled engineers can do with traditional camera designs. The need to have bulky lenses and the electronics that help them focus are the limiting factors in old-style systems, and while these lie at the heart of camera systems, size will always be an issue.
Thinking Outside of the Box
In order to drastically reduce the size of camera/lens systems, and also make a step increase in their performance, developers knew that they needed new thinking. This has led to research into the use of minute metasurfaces. Using a novel lens module design that integrates metasurfaces, the researchers have produced an ultra-thin camera system that lowers the thickness of a standard lens system by less than half, therefore addressing this restriction. On a pixel-by-pixel basis, metasurfaces may exactly adjust the three characteristics of light: intensity, phase, and polarisation.
A metasurface of this nature is composed of tiny nanostructures that are spaced apart by just a few hundred nanometres, making a small but technically stable assembly. According to the university team, by developing a metasurface optimised for a certain wavelength (typically 852 nm) and placing numerous sheets horizontally over a glass substrate, light may be reflected many times inside the substrate, resulting in space-efficient folded light routes.
From this the researchers demonstrated a construction for a miniaturised camera – called a metalens – that takes pictures using a series of thin, folding lens modules that alter the flow of light. This method not only overcomes the physical constraints of bulky traditional lens modules, but it also produces a much higher picture quality. This is due to its 10-degree field of vision inside a compact system footprint of only 0.7 mm thick, as well as its ability to produce high-resolution pictures near to the diffraction limit at an aperture of f/4 and a wavelength of 852 nm.
Further development work
Despite the advanced state of work on these new lenses, don’t expect to see this technology in a product any time soon. The manufacturing techniques need to be finalised, and there are some technical issues that need ironing out.
The metalens system is created using atomic layer deposition after being constructed using nanoimprint lithography, which is an efficient and scalable process that enables lenses to be mass-produced, once perfected. Even though it’s optimised for light focussing, the current design of metalens is prone to chromatic aberration and other distortions caused by light of different wavelengths interacting with it.
Researchers have found that the solution is to train an AI deep learning model to detect aberrations in the surface of the metalens and fix the blurring and colour distortions it causes. This method is novel because it corrects upcoming system pictures based on what it learns from a huge collection of photographs.
Two neural networks are trained in tandem using adversarial learning, which is employed by the Picture Restoration Framework. Much like the way in which AI systems are trained, the system is always being driven to improve as one network produces rectified photos and the other evaluates their quality. The model also learns that distortions vary with viewing angle with the use of sophisticated methods like positional embedding.
The technology is not quite there yet, but once it is, we are likely to see smaller, thinner, optics systems and from there, XR devices that are as small as a pair of glasses, and that is very exciting indeed.
Unity Developers are a specialist software house that is expert in creating games and Apps from first principles and making them a commercial success. If you have an idea for an exceptional game or App, contact us and let’s start developing.
