Changing Sketches into Photorealistic Masterpieces

A deep learning model developed by NVIDIA Research turns rough doodles into highly realistic scenes using generative adversarial networks (GANs). The tool is like a smart paintbrush, converting segmentation maps into lifelike images