Holo There! A New Camera Brings Holography to Life

This holographic camera captures and displays 3D scenes in real-time using a liquid lens and AI, making holography more practical than ever.

nickbild
9 months ago Displays
A novel, real-time holographic camera (📷: Z. Li et al.)

It seems as if we have been teased by the promise of holograms for decades now. These cool, three-dimensional displays occasionally turn up in the wild, as was the case with Sega’s 1991 arcade game Time Traveler, but only rarely. Most of the time, fake “holograms” in science fiction movies are as close as we are going to get to the real thing. But this technology has been around for a long time now — and there is no question that producing three-dimensional images has many applications — so where are they?

The problem is that, despite having been around for so long, many of the technologies required to produce a hologram have not advanced very much over the years. The sorts of technological booms that have been experienced in, say, artificial intelligence or computing, have not happened in this area. So producing high-quality holographic images is still an expensive, difficult, and time-consuming process, which makes the technology undesirable for most potential use cases.

An overview of the approach (📷: Z. Li et al.)

However, a new breakthrough may finally push holography into the mainstream. A team led by researchers at Beihang University has developed an innovative holographic camera that not only captures real 3D scenes in real-time, but also reconstructs them with exceptional accuracy. This development eliminates several long-standing obstacles in holography, such as slow capture speeds, low image fidelity, and incorrect depth blur.

The newly proposed holographic camera consists of two main components: an acquisition end and a calculation end. The acquisition end features a liquid camera equipped with a unique elastic membrane liquid lens, which is capable of rapid focus adjustment. Unlike traditional lenses, this liquid lens is actuated by a voice coil motor, allowing it to capture multiple focal planes within just 15 milliseconds.

A look at the hardware (📷: Z. Li et al.)

Traditional liquid lenses have been considered for similar applications in the past, but their slow response times (often around 100 milliseconds) made them unsuitable for real-time 3D capture. The new elastic membrane liquid lens overcomes this limitation by using an innovative control mechanism. When an electric current is applied, the Ampere force moves a press plate inside the lens, altering the shape of the elastic membrane and, consequently, its focal length. By precisely controlling this current, the lens can quickly shift between different focal planes, effectively capturing a full 3D scene almost instantly.

Once the multi-focal plane information is acquired, it is processed by a neural network called FS-Net, specifically designed for real-time hologram generation. Unlike conventional hologram computation methods, which rely on diffraction-based calculations that are highly time-consuming, FS-Net utilizes deep learning to rapidly translate the captured focal stack into a high-fidelity hologram.

An important feature of FS-Net is its use of Zernike polynomials to compensate for optical aberrations. This ensures that the final holographic reconstruction not only looks sharp but also maintains depth accuracy, eliminating the mismatch between the original 3D scene and its holographic projection. Additionally, the FS-Net training process integrates a focus stack renderer that simulates the blur and depth variations of real-world 3D scenes, improving the model’s ability to handle real-life visual complexity.

The FS-Net training method (📷: Z. Li et al.)

Using this approach, a complete, high-quality holographic image can be generated in just 13 milliseconds, making the total time from capture to display only 28 milliseconds — fast enough for real-time applications.

While further improvements and optimizations are still needed before holography becomes a common feature in consumer electronics, this new camera is a significant step toward making the technology a practical reality. With its combination of cutting-edge optics, high-speed processing, and deep learning algorithms, it finally feels like our wait may be nearing its end.

nickbild

R&D, creativity, and building the next big thing you never knew you wanted are my specialties.

Latest Articles