Manipulate images with a simple click: AI DragGan allows you to transform photos in record time

German and American researchers developed an innovative approach, called Drag Your GAN, that allows users to have flexible and precise control over the pose, shape, expression and layout of objects in still images. Despite still being on paper, the tool could change the way people edit their photographs.

Until then, existing methods for “animating” images required manually annotated training data or pre-existing 3D models. These approaches have shown limitations in terms of flexibility and accuracy. 

What is Generative Artificial Intelligence? |Newsverso Glossary

Generative Artificial Intelligence, or GAN, is a technique in which two neural networks work together to create new things. One of the networks creates new things, while the other tries to tell whether these things are real or fake. Over time, the network that creates new things learns to make things that look real, and the network that verifies learns to tell the difference between what is real and what is fake.

With DragGAN you can make someone smile in a photo just by dragging the mouse

DragGAN uses a less explored technique, allowing users to “drag” specific points in an image to achieve desired results in an interactive way. Just with your finger or mouse you can make someone smile in a photo.

This means we can create images of animals, cars, people and landscapes with personalized details, in a realistic way.

With DragGAN, anyone can warp an image and have precise control over the pose, shape, expression and layout of various objects. This flexibility allows for a wide variety of GAN-generated images, producing realistic results even for complex challenges such as hidden objects or deformed shapes that follow the object's rigidity.

It may seem scary. But if used ethically, this discovery promedrive the development of new applications in areas such as design, digital content creation and even games and films. You won't need to undergo complex processes in Photoshop, for example, to make an image the way you want.

With this advancement, the ability to control and manipulate images generated by artificial intelligence gains a new dimension, opening up fascinating possibilities for creating visually stunning and personalized content. 

The tool is not yet available to everyone. Only developers are testing the platform. However, the project has been registered on an open curated research sharing platform from Cornell University.

Testers also published previews on Twitter of how the tool will work when released generally. 

See also:

Artificial Intelligence: between revolution and regulation – professionals discuss the future of AI

The popularization of ChatGPT raised expectations, enthusiasm and fear in the world about the future of Artificial Intelligence. To what extent does technology come to add to modern society? And when does it pose a real risk to business and even the future of humanity? Experts and governments debate regulation and ethical brakes. We spoke to experts to understand what can/should be done in this regard. Follow 🧵

Midjourney releases new function that creates texts from images

O Midjourney launched a new feature that promeget involved with the art and design industry. It is an image-to-text conversion tool that allows users to upload an image and receive accurate and immediate text descriptions generated from it. Previously, the platform only allowed the opposite way, for the user to send a text command to receive an image from the prompt.