Otávio Good keeps a plaque with a few words of Spanish on the wall of his office in San Francisco so he can perform a magic trick for visitors. "Bienvenido al Futuro," it says. Good points the camera on his iPhone at the sign. The plaque appears on the phone's screen, and Good smiles as the Spanish words flicker and transform into English: "Welcome to the Future."
The mobile app powering this illusion is Word Lens, and it's the result of over two years of work by Good, a 36-year-old former video game coder who sold his startup Secret Level to SEGA for $15 million in 2006. His next business idea was born two years later, during a weeklong trip to Germany with his girlfriend. Good grew frustrated that there was no easy way for him to read German street signs, menus, and package labels. Paper dictionaries and electronic translators were too slow; online options meant incurring expensive cell phone roaming charges.
Word Lens relies on a series of complex computer algorithms to produce its presto change-o effect. The app employs optical character recognition, a technology commonly used in document scanners, to pick out letters from whatever the phone's camera is pointed at. It then matches words and phrases to a database of translations that Good assembled by comparing transcripts from sessions of the European Parliament, which are available in multiple languages. (Google used a similar technique for its Google Translate service.) The app places the translated text over the original—giving the effect, for example, that "hola" has been erased and replaced with "hello."
All these tricks occur in a split second, and no Internet or Wi-Fi connection is necessary. Scan your phone over a menu and it translates instantly. "Video game technology applied to the problem of language translation is what makes it fast," says Good. Some of the sophisticated methods that Word Lens uses, such as pixel shading and vector processing, are already common on game consoles and desktop PCs. Good mastered them during more than a decade working on complex video games such as SEGA's Iron Man series. Those techniques weren't possible on phones until Apple adopted a new, zippy graphics processor in the iPhone 3GS released in June 2009. By then, Good had already quit his job to work on Word Lens. "It was an educated bet," he says. "I crossed my fingers and hoped the technology would catch up to what I was thinking."
The program fits into an emerging category of apps known as augmented reality, in which cameras and mobile devices are used to layer virtual information onto the real world. While some of these apps feel like little more than parlor tricks, Word Lens may be the first that serves an everyday purpose, says Robert Scoble, a tech blogger. "It worked great at my local Mexican restaurant where the menu is in Spanish," he says.
Good released the Spanish version of Word Lens, which costs $10 per language, last December. He says the app's first few days of sales recouped the investment he'd already made, and he's looking to expand by hiring engineers versed in computer vision, the science of getting robots to see and interpret images. More languages will be available for the app later this year, starting with French. Good says he's turned away investment offers from venture capitalists and walked away from acquisition talks with big tech companies. "This technology has a lot of potential," he says. "And I want to build the company up to realize that potential."
SEGA bought Good's video game startup Secret Level for $15 million in 2006
He mastered high-end computing techniques while working on games such as Iron Man
To build his database of translations, Good relied on transcripts from European Parliament sessions