Compound eyes (CEs) offer a unique design paradigm for developing artificial micro-optical systems. However, current bionic CEs still fall significantly short of their insect counterparts in performance, particularly regarding ommatidial spatial arrangement, size distribution, visual field adaptability and environmental perception. Inspired by the CEs of Sympetrum frequens, we developed a tunable bionic CE with bi-focal ommatidia by integrating a flexible lens-on-lens (LoL) array with microfluidics, enabling simultaneous bi-focal imaging. The polydimethylsiloxane (PDMS) LoL array was fabricated through sequential processing steps, including femtosecond laser dual-modification of quartz glass, two-step wet etching, and soft lithography using the resulting glass template. Integration of this flexible LoL array with a microfluidic chip produced a tunable CE with bi-focal ommatidia, allowing liquid-pressure modulation of CE configurations. We systematically investigated the optical performance of both LoL ommatidia and the integrated CE, including bi-focal focusing/imaging properties, tunable field-of-view (FOV), large-FOV imaging, and moving target monitoring. This tunable CE with LoL ommatidia demonstrates significant potential for cutting-edge applications.

In video conferencing and facial recognition applications, eyeglass reflection often obscures critical facial features. Polarization information helps disentangle reflection and transmission components. Existing polarization reflection removal approaches rely on large-scale paired polarization datasets, making them hard to generalize to the scenes under unseen lighting conditions. To address these challenges, we propose PDPrior, an untrained polarization-guided diffusion prior for eyeglass reflection removal. The PDPrior requires no training data, and no ground-truth reflection-free images. It operates solely on polarization observations collected at test time to achieve artifact-free, high-fidelity reflection removal. The PDPrior leverages the generative prior of a diffusion model and introduces polarization as guidance to control the generation process. The self-supervised loss based on the forward model is employed to alternately update the reflection and transmission variables, endowing the generative model with physical interpretability. Extensive experimental results validate the effectiveness and robustness of the method on eyeglass reflection data captured in both indoor and outdoor lighting conditions, achieving higher face image quality assessment scores for recognition performance compared to state-of-the-art methods.

Infrared imaging is indispensable for its ability to penetrate obscurants and visualize thermal signatures, yet its practical use is hindered by the intrinsic limitations of conventional detectors. Nonlinear upconversion, which converts infrared light into the visible band, offers a promising pathway to address these challenges. Here, we demonstrate high-efficiency infrared upconversion imaging using nonlinear silicon metasurfaces. By strategically breaking in-plane symmetry, the metasurface supports a high-Q quasi-bound states in the continuum resonance, leading to strongly enhanced third-harmonic generation (THG) with a conversion efficiency of 3×10⁻⁵ at a pump intensity of 10 GW/cm². Through this THG process, the metasurface enables high-fidelity upconversion of arbitrary infrared images into the visible range, achieving a spatial resolution of ~6 μm as verified using a resolution target and various customized patterns. This work establishes a robust platform for efficient nonlinear conversion and imaging, highlighting the potential of CMOS-compatible silicon metasurfaces for high-performance infrared sensing applications with reduced system complexity.

The convergence of artificial intelligence (AI) and metaphotonics is creating a new paradigm for controlling light-matter interactions. The synergy of AI's ability to learn complex relationships in multidimensional data and provide ultra-fast inference with the capacity of metaphotonics to engineer optical properties not found in nature is unlocking a new era in computational design, real-time control, and fully automated optical systems. This review provides a comprehensive overview of state-of-the-art AI-driven approaches for metaphotonic systems. We focus on the solutions to real-world problems in accelerating metaphotonic simulations and inverse design, optical data characterization, and the development of fully integrated end-to-end AI-assisted metaphotonic systems. Finally, we provide our perspectives on the future research directions and emerging opportunities at the rapidly evolving intersection of metaphotonics and AI.

Metasurface fabrication still faces critical processing challenges in balancing the structural order and disorder, achieving high-speed patterning, and extending material compatibility to refractory metals for operation under extreme conditions. This study demonstrates that femtosecond laser maskless direct writing (fs-LMDW) offers a versatile platform for engineering multispectral information, all-in-one metasurfaces on pure zirconium (Zr) substrates. Through sequential fs-LMDW in air and ethylene glycol (EG), deceptive grey-colored visible information is superimposedly encoded with the infrared (IR)-encrypted information (invisible among black-color structured background), which leverages the crosstalk-free structural modulation of singular-band IR and visible light. The metasurface exhibits robust thermal stability and high-security encryption capability across a wide temperature range, with IR-concealed information (such as QR code) remaining securely encrypted until thermally activated at 300 °C for smartphone-readable information decryption. Furthermore, the visible information is both erasable through 300 °C oxidation heating in air and rewritable via fs-LMDW in EG without compromising IR encryption security. Particularly, the one-time complete erasability makes it possible to identify whether the encrypted IR-information has been decrypted, underscoring the robustness and high security of the platform. The presented hierarchical micro/nanostructuring methodology is deemed to be applicable to a large material matrix to gain high-security, and multifunctional metasurfaces that are more difficult and complex for current metasurface fabrication techniques.

Scene-level high-precision 3D imaging remians a key challenge due to the trade-offs between imaging distance and accuracy. While polarization-based 3D imaging offers potential, it offen fails facing "discontinuous" targets—environments where multiple objects are separated by space resulting from surface normal integration. A recent work integrating binocular stereo vision with polarization information successfully recovers depth by iteration. This intergation-free appoach paves the way for scene-level high-precision 3D imaging and future applications.