If the name DragGAN still doesn't sound familiar to you, get ready for a radical change in the world of photo editing. The old way of editing images, with complex tools and endless Photoshop layers, is about to give way to something completely new. Literally enthralling, if you'll pass the term.
A new “tremendous toy” for creatives
The idea of DragGAN was expressed in a paper (I link it here) by a mixed team of researchers between MIT, Google and the University of Pennsylvania. It is to bring simplicity to a new level in image editing. It is a solution that does without all the classic photo editing tools: brushes, layers and everything else, which have made the fortune of programs like Photoshop.
The user can simply click at strategic points on the photo, creating “intention” points, and then drag these points to modify the image. It's almost magic.
Photo editing enters the fourth dimension
What does the ability to edit an element in a photo mean in layman's terms (in the world where the photo it will already be "computational" in its own right) with a simple drag of the mouse? For example, you can open a person's closed eyes, lengthen a skirt, zoom in or out on a car, all while maintaining a realistic look and without having to make any manual additions to the image.
The enthusiasm generated by this innovation is palpable. After the announcement of its development, users went haywire the DragGAN website: a clear signal of the interest it is arousing in the world of photo editing, and more generally in image editing.
DragGAN is a clean break with the past
The most loyal readers of Futuro Prossimo may remember when we said to each other: le. GANs (generative adversarial networks) will be able to “imagine”, e.g they will make havoc. Now it happened, and again you might ask yourself: but how does DragGAN differ from other photo editing tools, or from generative AIs like Midjourney or StableDiffusion (which can also be used in combination)?
The answer is simple: DragGAN does not generate images. It modifies them. Literally. And it does so in a surprisingly effective way: if you look at two photos, one before and one after editing with DragGAN, you won't be able to tell which of the two is the original.
DragGAN is capable of doing things that no photo editing software in the world can do yet, such as changing the angle of an object, not just its perspective, or “inventing” the details needed to make everything more realistic.
A transformer. And it's just a tool still under research. I dare not think what he will be able to do when he is ready for mass.
The future of photo editing
Imagine a world where to edit an image you just need to give a voice command, without needing any specific skills. DragGAN is the first piece of a puzzle that will lead to such a future.
Think about it: “Hey Google, turn this shirt into a red dress and make the model wear a hat.” It may be the norm in the future, no more mouse or keyboard. Taking far fewer photos.
Sometimes none.