3D Photo Editing Could Make it Impossible to Tell If Images Are Real

Photographs are time's static records, whereas cinema is, it's been said, is " sculpting in time." One captures a singular moment, while the other simulates life, allowing the tucked away corridors both in and outside the frame to be revealed over time.

But now,  researchers at Carnegie Mellon and University of California, Berkeley are giving the world of still photograph a piece of photo-editing software that will allow users to create 3D manipulations to illuminate the hidden areas of an image.

The 3D-photo editing software works by ironing out the graphic flaws in an object's lighting, texture, and shape. It does this using a wealth of images of everyday objects uploaded to the internet on a daily basis. It's image manipulation with the help of Big Data, and the realistic accuracy puts today's Photoshopping to shame.

This could result, researchers suggest, in photo editors being able to convincingly turn cars over, make grandma do a backflip, fake a baby lifting a heavy sofa, make a paper-crane flap its wings (as shown in the video), or manipulate airplanes in a historical photograph to "change its story."

But beyond that, there's a glaring potential dark side to this software if it's misused. In the wrong hands, all sorts of deceptive images could hit the internet. Anyone who really wanted to control an image would suddenly be able to do so in a very convincing fashion.

That however, is not the researchers' intention. The idea is to enable users to manipulate the photograph while maintaining realism. Professional photo editors, and maybe even future Instagram users, could also use this type of software to effectively drop 3D models of new objects into the scene, achieving a sort of fantastical surrealism in the process. Artists, maybe even the masses, would likely relish this sort of creative story-bending power.

As researchers demonstrate in the above video, the software permits users to move photographic elements in 3D fashion, letting users see what lies beneath, behind, below, and in various other locations within the photograph usually obscured during exposure.

By using pixel information in the visible parts of objects, it's possible to overcome three types of graphic mismatch that exist between photographed objects and stock 3D modeling: geometry mismatch (the object's shape, whether manmade or natural), appearance mismatch (texture and surface reflectance), and illumination (lighting effects).

In other words, photo editors will be able to leverage Big Data's repository of stock 3D images to manipulate photographs at will.

"Photo-editing software restricts the control of objects in a photograph to the 2D image plane," the researchers wrote. "We present a method that enables users to perform the full range of 3D manipulations, including scaling, rotation, translation, and nonrigid deformations, to an object in a photograph."

As the researchers noted, this type of effect is impossible in traditional 2D photo-editing programs. With this new 3D-based software, photo editors could engage in some surreal and magical special effects. As they said in their research paper, it's all part of encouraging people to become active creators instead of "passive consumers of visual information."

Ultimately, they want users to be able to manipulate "what we know about the scene behind the photograph."

The researchers want creators to be able to change a photography's story. If this works as they imagine it will, they're going to get that and then some. Let's hope that this photo manipulation software is used for worthwhile creative enterprise instead of deception.