Abstract
We propose, WarpGAN, a fully automatic network that can generate caricaturesgiven an input face photo. Besides transferring rich texture styles, WarpGANlearns to automatically predict a set of control points that can warp the photointo a caricature, while preserving identity. We introduce anidentity-preserving adversarial loss that aids the discriminator to distinguishbetween different subjects. Moreover, WarpGAN allows customization of thegenerated caricatures by controlling the exaggeration extent and the visualstyles. Experimental results on a public domain dataset, WebCaricature, showthat WarpGAN is capable of generating a diverse set of caricatures whilepreserving the identities. Five caricature experts suggest that caricaturesgenerated by WarpGAN are visually similar to hand-drawn ones and only prominentfacial features are exaggerated.