AI image editing technology DragGAN came out, turning elephant P around just by dragging the mouse

Thanks to CTOnews.com netizen Sancu for the clue delivery! CTOnews.com, May 21 / PRNewswire-FirstCall-Asianet /-- researchers from the Max Planck Institute of computer Science, MIT CSAIL and Google have developed a new way to control GAN, DragGAN, which allows users to "change the size of the car or turn a character's smile into a frown" by clicking and dragging. DragGAN is still a research paper, but it has attracted a lot of attention because so many people watched its demo video that the research team's home page crashed due to too much traffic.

DragGAN is similar to the Warp tool in Photoshop, but is more powerful because instead of "squeezing pixels", it "regenerates objects" and can even rotate 3D images. The potential of this tool is that artificial intelligence that generates an image from text does not always output the results you want, so you can edit the image after it has been generated without having to generate a new one. The research paper shows a number of demonstration cases, including increasing the height of the mountain, changing the posture of the model and the length and shape of the clothes, opening or closing the lion's mouth, and changing a person's expression from insipid to smiling.

At present, many companies are trying to provide some features for generative AI content, but they cannot edit AI-generated images.

CTOnews.com noted that since DragGAN is currently just a demonstration, it is not clear when it will actually be launched and how it will work, especially given that the demo videos are based on low-resolution images. However, the potential of this technology is still very great, if you also receive a P-map request to "turn the elephant around", this technology can help you complete the task in the future.