UCLA Engineers Project 28-Layer 3D Images in a Single Shot Using AI-Optimized Light
A new system from UCLA engineers projects 28 distinct image layers in 3D space simultaneously, using AI to encode and decode light patterns. This breakthrough could enable ultra-compact volumetric displays for medical imaging and augmented reality.
Imagine a hologram so precise it can project 28 layers of 3D imagery in a single snapshot, without bulky lenses or moving parts. UCLA researchers have achieved exactly that by merging artificial intelligence with optical physics, creating a system that could shrink theater-sized projection tech into smartphones. Their light-programmed encoder-decoder system, detailed in Science Advances, marks the deepest multi-plane 3D projection ever demonstrated in one exposure—potentially transforming medical scans, AR headsets, and even 3D printing.
- Projects 28 image layers across 1.5 cm depth with 92% fidelity
- Uses passive diffractive optics (no power needed) paired with an AI-trained encoder
- First system to combine deep learning with physical light manipulation end-to-end
- Enables real-time 3D visualization without glasses or tracking
What Happened
The UCLA team, led by bioengineering professor Aydogan Ozcan, designed a hybrid digital-physical system where a spatial light modulator (a pixelated device that shapes light waves) encodes 2D images into specific light patterns. A 3D-printed polymer plate, its surface etched with nanoscale features learned by AI, then decodes these patterns into 28 discrete focal planes. In tests, the system reconstructed a 3D model of the Stanford Bunny—a standard graphics benchmark—with layers spaced 0.5 mm apart. Unlike prior volumetric displays needing mechanical scanning or multiple projectors, this approach requires only a single light source and static optics. “We’ve effectively turned a slab of plastic into a computational lens,” said Ozcan.
The Bigger Picture
This technology could democratize 3D visualization, replacing MRI workstations costing $250,000 with tablet-sized devices. Surgeons might soon rotate CT scans mid-operation without touchscreens, while engineers could inspect prototype parts in mid-air. The passive decoder’s low power needs make it ideal for AR glasses—a market forecasted to hit $88 billion by 2030. “Most volumetric displays trade resolution for depth or vice versa,” said Gordon Wetzstein, Stanford electrical engineering professor unaffiliated with the study. “This work elegantly sidesteps that compromise using machine learning.” Challenges remain in scaling the decoder for larger displays, but the team has filed patents for consumer and medical applications.
What Comes Next
Ozcan’s group aims to double the layer count within two years while shrinking the encoder to fit mobile devices. Major hurdles include extending the projection depth beyond 2 cm and adapting the system for video rates (currently static images only). Toshiba and Siemens Healthineers have expressed interest in licensing the tech for diagnostic imaging. Consumer prototypes could emerge by 2026—potentially enabling glasses-free 3D smartphones or compact AR projectors priced under $500. For now, the research provides a blueprint for merging AI with photonics, a combination that could redefine how we interact with digital information.
Q: How is this different from holograms?
Traditional holograms create the illusion of depth using light interference, while this system projects physically distinct layers—like transparent slides stacked in space—enabling true volumetric viewing from any angle.
Q: Could this replace VR headsets?
Not immediately; current prototypes lack the speed for interactive VR, but the tech could enable thinner AR glasses by eliminating bulky waveguide displays.
