Fig. 1: A schematic of geometric image transformation using metasurfaces.

The image of an object is projected on a subwavelength-thin metasurface with nanoscale spatial resolution and is then converted into an intentionally distorted image by repositioning pixels in a constrained manner. The light (the green dotted line) transmitted from each pixel of the image with amplitude-only transmittance \(f\left(x,\, y\right)\) in the (x, y) plane is directed by the metasurface with a phase profile \(\varphi \left(x,\, y\right)\) to (X, Y) plane, forming an image with a redistributed intensity profile \(g\left(X,\, Y\right)\). u and v denote arbitrary pixels in (x, y) and (X, Y) planes, respectively. f is the distance between the input and output planes.