It might seem like a witchcraft, but Nvidia researchers have developed an advanced image retouching tool that can intelligently reconstruct incomplete photos.
Removing unwanted artifacts in image editing is nothing new: Adobe Photoshop's Content-Aware tools are pretty much the industry standard: the prototype tool that Nvidia is showing looks incredibly impressive.
Do not trust our word: check the two-minute video below to see what this new technology is capable of.
What differentiates the new Nvidia tool from something like Content-Aware Fill in Photoshop is that it analyzes the image and understands how the subject should look; Content-Aware Fill is based on the surrounding parts of the image to complete what you think should be there.
The Nvidia tool is a much more sophisticated solution. For example, when trying to fill a hole where an eye would be in a portrait, as well as using information from the surrounding area, Nvidia's deep learning tool knows that an eye must be there and can fill the gap with a realistic alternative generated by computer .
"Our model can robustly handle holes of any shape, size, location or distance from the edges of the image," the researchers write. "Previous approaches to deep learning have focused on rectangular regions located around the center of the image, and often depend on costly post-processing, and our model graciously handles holes of increasing size."
At the moment there are at least no words when we are likely to see this tool more widely available. To know, it gives us an idea of the near future of image editing.
