What is DragGan? How to use it?

A group of researchers from Google along with the Max Planck Institute of Informatics and MIT CSAIL, have developed DragGAN, (Generative adversarial networks) an intuitive point-based image editor.

What is DragGan?

DragGan is an AI Tool that lets the users manipulate the image and change the pose, shape, landscape and even the emotions of the subjects in the picture by just Dragging and dropping points on the image. This tool uses comprehensive Artificial intelligence data and it is still in the development phase.

How does it work?

This new technique for interactive point-based manipulation of images, Drag Your GAN, uses a generative image manifold to learn all the possible images that a GAN can generate. This manifold consists of a high-dimensional space that contains all the possible images that the GAN can generate.

After the learning of manifold, Drag Your GAN enables users to deform a picture by dragging points on it. The movement of the points toward their target positions on the manifold results in a new image.

Source: Max – Planck Gesellschaft

GAN uses two neural networks—generator and discriminator. They are set against each other to generate new and synthesized instances of data. But diffusion models are likelihood-based models that offer more stability as well as higher quality when creating images.

Features of Drag your GAN:

  1. The interactive point-based manipulation technique is one of its new feature.
  2. Its mask feature enables you to make changes in the specific region of the image.
  3. Drag Your GAN enables users to deform images with precise control over where pixels go. This contrasts with traditional image editing techniques, which often require large, imprecise changes.
  4. This makes it a versatile tool for a wide range of image editing tasks, including changing a picture’s pose, shape, expression, and layout.
  5. A reliable tool for image editing tasks that require high accuracy. It produces realistic results even in challenging scenarios.

How to use DragGan AI for Photo Editing?

Follow the below given steps to edit your photos using DragGan AI

What you can do with DragGan tool?

DragGAN is a versatile image editing tool. Although it is still in its early stages of development, it has already shown remarkable possibilities. It is exciting to watch how DragGAN progresses in the future.

Lets look into some of the things you can try with DragGan

1 . Change the Pose of a Person:

DragGAN makes it simple to change the pose of people or animals in an image. You can modify their placements, angles, and orientations to achieve desired or different stances.

2 . Change the shape or size of an object:

DragGAN allows you to manipulate objects within a photo. Objects can be stretched, reduced, or twisted to achieve interesting results.

3 . Make changes in a face :

You can make specific changes in an image of a face of a person by selecting specific area of the image. For Example, you can select part of the image where eyes are located and make changes to only that region of the image. You can refer to the video below.

What types of images can be manipulated with DragGAN?

DragGAN supports a wide range of image categories, including animals, autos, people, landscapes, and others. Users can alter the image’s layout, as well as its placements, shapes, emotions, and other elements.

How to use Mask feature in DragGan?

You can use a feature called Mask. Users can use it to select the specific region of the image and make manipulation to that part of the image only. You can checkout the usecase of mask feature in the above image.

Conclusion

While diffusion models for image synthesis are gaining popularity, general adversarial networks (GANs) also gained popularity and sparked interest.

According to this new technique, GAN models are more powerful than diffusion models used in tools like DALLE.2, Stable Diffusion, and Midjourney.

Follow us on InstagramFacebook, and Twitter!

Read on NFT- token Gating