An Optically Transparent Transmissive Metasurface for User‐Defined Broadband Electromagnetic Illusion Generation

Electromagnetic illusion technology offers a disruptive approach to advanced camouflage by precisely tailoring the object's scattering signature. Over recent decades, the emergence of metamaterials and metasurfaces has revolutionized this field. However, conventional electromagnetic illusion devices typically employ reflective arrays to reconstruct electromagnetic field distributions for basic demonstrations, resulting in limitations such as feed blockage, optical occlusion, and prohibitive computational costs, which are further exacerbated in broadband applications. Here, an optically transparent transmissive metasurface (TTM), strategically designed using conditional generative adversarial networks, is proposed, which can not only generate user‐defined broadband electromagnetic illusions closely resembling specified targets but also significantly shorten the inverse design cycle compared to traditional iterative optimization. Full‐wave simulations and microwave measurements confirm the proposed TTM's broadband robustness and immunity to feed blockage and visual obstruction. The prototype exhibits a light transmittance of 71.5% and produces high‐fidelity illusions across 12 to 18 GHz, with forward‐scattering radar‐cross‐section patterns maintaining >90% correlation with the desired profiles. By integrating transparent photonic engineering with deep‐learning‐assisted inverse design, this work establishes a versatile platform for intelligent electromagnetic manipulation and paves the way for adaptive camouflage systems that reconcile optical transparency with advanced radio‐frequency deception.