Blog

Copy & paste the real world with this new AR app

Augmented Reality and Artificial Intelligence have made incredible progress in recent years. Now there is a new, exciting use-case for AR. Cyril Diagne, developer at Google Arts and Culture, demonstrated in a Twitter video a new augmented reality tool Cut & Paste that has the ability to copy and paste real-life objects into the digital world.

The video shows how Diagne takes a snap of various real objects and then adds them into a Photoshop project. In his thread, he explains that a first component separates the foreground object from the background with the help of machine learning. Then a second component detects where the phone is pointing at on the PC screen. The whole process takes around 6.5 seconds to complete: 2.5 seconds to copy the object and four seconds to paste it. Currently the tool requires Photoshop as an output. However, Diagne notes that it might support different outputs in the future.

Promising use-cases

AR Cut and Paste is definitely impressive as the tool reverses the AR process as we know and use it to date. So far, other applications use AR to project digital images into the real world. Especially in the area of shopping and advertising this technology is already frequently used. For example, to show us what clothes, make-up, furniture and other objects would look like on us or into our surroundings. AR Cut and Paste reverses the process by bringing real objects into the digital world.

Looking into the future, this technology could be perfect for quickly capturing visuals from the real world and inserting them into digital documents. A useful feature for a wide variety of applications, such as presentations, designs or also image editing. Instead of taking a photo, editing it and then inserting the cut-out into wherever it is needed, simply point the smartphone camera at the object and copy-paste it to the PC. Which speeds up and simplifies the process considerably.

Diagne points out, that the tool is currently a research prototype and not a commercial user tool. Replies to the video show, however, that apparently a few other companies are already working on similar software. Diagne himself has published his code on GitHub for anyone to check it out and help to improve the tool. So we can be excited when this technology will take the next big step.