Researchers have demonstrated an experimental AI tool that can fully process images in no time. For example, it should be possible to rotate a face and adjust facial expressions within two mouse clicks.

The tool, DragGAN, aims to make specific changes to specific parts of images without changing the entire image. This requires users to indicate two points on the image; The part to be manipulated and the place in the image where that part should go. The tool should do the rest, say the researchers.

DragGAN allows anyone to distort an image with precise control over where the pixels go, manipulating the positioning, shape, expression, and layout of diverse classes such as animals, cars, people, and landscapes. Unlike, say, the Warp tool in Photoshop, existing pixels don’t change, but the neural network creates completely new pixels.

The examples shown show, among other things, that the tool can open a lion’s mouth and then make the animal look in a different direction by turning its head. People’s clothes, such as the sleeves, are also lengthened, and the arms and legs can be seen moving. Another example shows how landscapes are fully modified.

Most of the images in the examples are generated by artificial intelligence, but according to the researchers, the tool should also be able to handle real images. This is evidenced by a smiling image of US President Joe Biden, who in a few minutes changed to a shocked facial expression, with his head also turned.

According to the researchers, the tool is capable of achieving “realistic results that continuously track object hardness, even for challenging scenarios, such as hidden parts.” The team prepares To publish the tool code to GitHub next month.

Six researchers working at German and American universities contributed to the project. Abhimitra Mika, Research Scientist in Google’s AR division, also contributed.