DragGAN AI Tool Lets You Click And Drag To Manipulate Images, And It’s Wild
![hero draggan ai software demo](https://images.hothardware.com/contentimages/newsitem/61664/content/hero-draggan-ai-software-demo.jpg)
Have you ever wanted to give grumpy old Uncle Dave a smile in a photo you snapped of him on Christmas morning, or change the position of your hyperactive dog that is impossible to snag a good photo of? Well then, DragGAN may be the editing tool you are looking for. It is similar to Adobe's Photoshop Warp tool, but takes the ability to change an image to the next level. Instead of simply being able to stretch pixels in a funny and creative way, the new AI tool is able to regenerate areas of the image that did not exist before. As of right now, examples of the AI-driven software are only available via demos online at the moment, but show a ton of potential.
Researchers point out that while current software is able to gain controllability of generative adversarial networks (GANs) through manually annotated training data or prior 3D models, the new approach explores a new way of manipulating GANs. It does this through two main components. The first is a feature-based motion supervision that drives the handle point to move toward the target position. The second involves a new tracking approach that takes advantage of the discriminative GAN AI features in order to keep localizing the position of the handle points.
The research paper states that "anyone can 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."
While the concept is still in development by the team, it shows a lot of promise. The thought of being able to "make a picture of a lion stalking through the savannah" is intriguing, to say the least. The researchers remarked that their "approach can hallucinate occluded content, like the teeth inside a lion's mouth, and can deform following the object's rigidity, like the bending of a horse leg."
More demos of DragGan are available to view via the online publication of the research paper.