This artificial intelligence creates the movements you want on the photo

In the world of artificial intelligence specializing in image generation, here’s a newcomer. its name Dragan Its goal is not to create an image from words, but rather to reorient the image and modify certain elements of it by artificially regenerating them with a few mouse movements. With it, it is possible to make you smile back in a photo, for example, or to make a cat wink and a lion turn its head. In principle, the utility is reminiscent of Photoshop Warp from Adobe, but with a significant plus. Instead of stretching pixels with the mouse, fully integrated AI technology will create non-existent areas of the image. To design this amazing tool, researchers from major institutions and companies, such as Google, MIT, and the University of Pennsylvania, worked together. They have published a page where you can learn how to manipulate the image and let the AI ​​edit. DragGAN’s GAN is an acronym for Generative Adversarial Networks. With them, the first network generates an image and its opponent chases down the wrong details.

competing networks

To fool the discriminator, the generative network must ensure that the result is of high quality and very realistic. In the case of this research working with DragGANs, the researchers have come up with a new way to deal with these GANs. The user will click on the part to be modified, to create an anchor, and then perform a movement allowing that anchor to be moved. GANs will then monitor these movements. but that is not all. The discriminating network will follow and perform critical analysis of this anchor points as they shift. Through this process, it is possible to distort an image with precise and realistic control over the new location of the pixels. Thus, a dog depicted frontally can be repositioned on three-quarters in a realistic manner. With this principle, the researchers show that artificial intelligence is able to invent what is not in the image. This is the case of the teeth of an open-mouthed lion, for example.