AI technologies have been rapidly advancing, and in a latest
development, allowing people to change the appearances of photos in just
seconds. From re-generating an underlying object to rotating images as if they
were 3D, you can manipulate images with a simple motion of your fingers, thus producing outputs close to your preferences.
“DragGAN” is a research project which lets a user “deform an
image with precise control over where pixels go, thus manipulating the pose,
shape, expression, and layout of diverse categories such as animals, cars,
humans, landscapes, etc.”
As described by the research team itself, the work involves
studying “a powerful yet much less explored way of controlling GANs, that is,
to ‘drag’ any points of the image to precisely reach target points in a
user-interactive manner.”
It is amusing to see that you can make 2D images appear in a 3D form by adjusting reflections on them, change the direction of a face, turn a smile into a frown, and so much more.
While previous AI models have also been able to generate
realistic images, DragGAN seems to beat them in terms of flexibility and
precision. Instead of merely smudging pixels, the model enables one to control the
pose, shape, expression, and layout of the generated objects. In other words,
you can create your image to resemble closely how you imagine it to be.
Below are some examples of how DragGAN works:
