Prototype iPad app offers Blade Runner-esque voice-controlled image editing


posted Wednesday, February 13, 2013 at 2:46 AM EST

"A picture," so the old saw goes, "is worth a thousand words." That may well be so, but if researchers at the University of Michigan and Adobe Research get their way, a few sentences could soon be sufficient to make that picture just a little bit more worthy. Their creation, a prototypical, voice-activated image editor called PixelTone, lets you make significant changes to the look and feel of a photo with just a few spoken words, perhaps combined with a touch gesture or two.

Anybody who's spent much time massaging their images in powerful image editing applications like Adobe's Photoshop and its key rivals knows just how tough they can be to navigate, even once you're familiar with the interface. For the casual user, even the friendlier consumer image editing options can prove something of a handful. 

That's where PixelTone comes in. The research team took a close look at how professionals and novices alike put into words the changes they wanted to make to their images, and set about creating an application that could understand and act upon those requests. The result is something that's more than a little reminiscent of the Esper machine concept created by industrial designer Syd Mead for the 1982 cult sci-fi movie Blade Runner, but with one important difference -- it's actually real. 

Even in prototype form on an iPad tablet, PixelTone looks mighty impressive, providing rich control over image editing with just the spoken word and a few touch gestures.

Although PixelTone lacks its fictional counterpart's surprising ability to navigate photos in three dimensions, the results shown in the short two-minute video demonstration above are nonetheless very impressive. Running on an iPad tablet using Apple's iOS 6 operating system, the app allows for a total of 21 different image processing operations. Some require no interaction beyond the spoken word, such as a filter effect that is applied simply by saying "Make it retro." Others combine a vocal command to select the tool needed, and a touch gesture to indicate how the tool should be applied. For example, you can "draw" crop marks on the screen and then command that the image be cropped, or request a midtone levels adjustment at the top of the image using speech, and then drag a slider with your fingertip to control the strength of the adjustment. Perhaps most impressively, you can identify specific faces or objects in the scene with a touch, then tag them with a spoken name. These individual objects can then be targeted directly by name in subsequent operations.

Of course PixelTone is, so far, still just a prototype. Although the research team has apparently completed a real-world evaluation suggesting that users already found the design effective and useful, there hasn't as yet been any suggestion that the app is ready for retail, nor that its underlying technologies might be headed to a computer or smart device near you any time soon. For consumers, though, there's no denying that the concept offers an interesting (and arguably, friendlier) alternative to the vast array of obscure buttons, controls, and deeply-nested menus typical of many current image editors. And even more experienced photographers who're better-versed in their image editor of choice might well find an additional control option attractive, if only to avoid a little carpal tunnel-inducing activity on mouse and keyboard.

In fact, the biggest drawback we can see to PixelTone is that, just 30 years after Blade Runner's debut, it makes one of that film's most iconic creations seem... well, altogether too realistic. For fans of that great movie, the news that Rick Deckard's futuristic photo workstation is almost a reality may be welcomed with just a hint of sadness. For the rest of us, though, PixelTone promises great things to come!

(via John Nack on Adobe)