New AI research allows users to manipulate images in seconds as well as clicking and dragging them. It is that simple. The new feature works like a Photoshop warp tool, but this one is far more powerful.
With this feature, you are not just smushing pixels around the place, but instead, you are using AI to re-generate the underlying object. You can even get to rotate images as if they were in 3D.
New AI Research Manipulate Images
No, it’s not quite over yet in regards to the ability of AI tools to manipulate images continues to grow. And the latest instance is only a research paper for the time being, but however, a very impressive one, allowing users to simply drag elements of a picture to change their appearance.
This however does not sound too exciting on the face of it, but you should take a look at the examples that have been shared to get an idea of what this system can actually do.
Not only can is it that you change the dimensions of a car or even get to manipulate a smile into a frown with just a simple click and drag, but you can also rotate the subject of a picture as if it were a 3D model thus changing the direction that someone is facing, for instance. One demo in question even shows the user adjusting the reflections on a lake as well as the height of a mountain range with just a couple of clicks.
Videos Have Been Shared Regarding the AI Research
Videos have been shared from the research team’s homepage, though they have been crashing due to the amount of traffic that has been sent to the site by Twitter (mainly by user @_akhaliq, who did quite a fantastic job in highlighting interesting AI papers and it is well worth a follow if that interests you). You also can read the research paper on arXiv right here if you wish.
And as noted by the team responsible, what is really interesting about this work is not necessarily the image manipulation per se, but however the user interfaces. We have been able to use AI tools such as GANs in generating realistic images for a while now, but many of the methods lack flexibility and precision. You can tell an AI image generator to simply “make a picture of a lion stalking through the savannah,” and you will get one, but it might not be the exact pose that you want or need.
DragGAN AI Tool
This very model, dubbed DragGAN, offers users a clear solution to this. The interface in question is exactly the same as traditional image-warping, but rather than simply just smudging and mushing existing pixels, the model in question generates the subject anew. And as the researchers write: “[O]ur 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.”
This Is a Demo Tool for Now
This obviously is just a demo for the time being, and it is impossible to evaluate the tech completely. (How realistic are the end images, for example, one may ask? It’s quite hard to say for now based on the low res videos that are available.) But it’s yet another example of making image manipulation more accessible.
MORE RELATED POSTS
