A team at UCLA has built a system that can project 28 separate image layers in three dimensions using a single snapshot of light. The work, published in the journal Light: Science and Applications, could push forward the development of compact, high-fidelity 3D displays for augmented reality, virtual reality, and holographic applications.
According to Phys.org, the research was led by Professor Aydogan Ozcan at the UCLA Samueli School of Engineering and the California NanoSystems Institute. The central challenge the team set out to solve is a longstanding problem in holographic display technology: when image planes in a 3D display are placed very close together, light spills between layers and degrades the image. That effect is called diffraction-induced crosstalk.
The UCLA system addresses this by combining two components trained together using deep learning. The first is a digital encoder built around a Fourier-based neural network. It takes a full stack of target images, factors in the depth position of each one, and compresses all that information into a single phase pattern. That pattern is what gets sent to the display hardware in one shot.
The second component is a physical decoder made of multiple layered surfaces that have been structurally optimized. As light carrying the encoded pattern travels through those surfaces, the surfaces steer each image to its correct depth while blocking light from leaking into neighboring layers. No active electronics are involved on the decoder side.
In computer simulations, the researchers demonstrated that the system could handle 28 axial image slices encoded into a single phase pattern, with layer separations on the order of a single wavelength of light. That is an extremely tight spacing. The simulations also mapped out how design choices, such as the number of diffractive layers, the resolution of the spatial light modulator, and the density of depth encoding, affect system performance.
The team then built a physical prototype to verify the approach. The experimental setup used a single-layer decoder operating in the visible spectrum and projected two image planes simultaneously. The measured light patterns matched both the simulations and the target images closely and performed better than a comparison system that used no diffractive decoder at all.
The researchers say the results confirm that the hybrid digital-optical architecture works in practice, not just in simulation. The design guidelines they developed are meant to help other engineers build future versions of diffractive 3D displays. Potential applications include next-generation AR and VR headsets, medical imaging systems, and compact holographic projectors where size and image quality both matter.
