Metaoptics for information processing

All-optical image processing enables ultrafast, energy-efficient computation by exploiting the parallelism of light. However, existing approaches often rely on bulky optics or are limited to single operations. We present a metasurface-based platform—meta-operators—that performs arbitrary image transformations in a compact, passive format. By integrating double-phase encoding and polarization multiplexing, we realize diverse analog operations including first- and second-order differentiation, cross-correlation, and high-fidelity 3D holography, all at visible wavelengths. This scalable design establishes a versatile framework for real-time optical computing, intelligent vision systems, and compact volumetric displays. https://arxiv.org/abs/2503.12252