Word Lens

» see questvisual.com as it lived, on the wayback machine

~ the last words ~

word lens's peak years overlap with my early career in turkey, but the cultural moment was an iphone viral demo cycle on american tech blogs in late 2010 and early 2011. i was on a feature phone for most of '10 and '11 and not in the demographic the app was selling to (close enough). anyway, the app launched on december 16 2010 as an iphone app from quest visual, a small studio fronted by otavio good. the first language pair was english to spanish for ten dollars per direction. you pointed the camera at a sign or a menu and the screen showed the same image with the words replaced.

the engineering pitch was that the translation was offline. no server round trip, no data plan needed, which mattered in 2011 because the audience was supposed to be tourists in foreign countries paying roaming rates. either way, the overlay preserved approximate font and position so the menu still looked like a menu. it was the first augmented reality demo that did something useful instead of placing a virtual model on a coffee table. the youtube demo videos went viral in the same news cycle as the early iphone 4 hype.

by 2014 quest visual had added five more languages on top of spanish. on may 16 '14 google announced the acquisition for an undisclosed sum. honestly, the deal terms made every existing language pack free for download until january 2015. for some reason, on january 14 2015 google released the new google translate app with word lens features built in as a camera mode, covering more languages and using google's cloud translation for the parsing.

the standalone word lens app stayed available for a transition window in early 2015. then it quietly disappeared from the app store later that year. for whatever reason, the quest visual website went down on the same schedule. the technology is what survived. the brand and the standalone product did not.

~ the record ~

~ how it landed ~

the first quest visual youtube demo went up in december 2010. it showed a phone pointed at a spanish street sign with the word "abierto" turning into "open" in roughly the same font on the screen behind the glass. that video crossed a million views in days and the iphone app was a top paid utility in the us app store before christmas. in those days, the demo was so unfamiliar that several tech reporters opened their coverage by guessing it had to be faked.

the second viral wave came in 2011 when language packs for french and italian shipped. the app worked offline with the language pack downloaded, which made it credible as a travel utility. anyway, the price was around five to ten dollars per language direction. by mid 2012 the lineup covered english to spanish and french in the first wave. italian and german came next, then portuguese and russian rounded the set out.

~ what it actually did ~

the underlying engine was ocr on the camera feed plus an offline dictionary, with a rendering pass that drew the translated text back onto the frame in roughly the same position as the source word. honestly, the translation was word-by-word rather than full grammar. you could read a menu but not a magazine article. the app could not handle handwriting or stylized fonts. it worked best on printed signs in plain sans-serif fonts under good light. in fairness, that covered most tourist use cases for the price.

~ what got buried with it ~

the google translate version added many more languages and the cloud kept improving. the offline magic of the original word lens, the part where it ran with no signal in a foreign country, did not fully survive the merger at first. google translate camera mode initially required a connection for the parse on most pairs. either way, by 2017 google had shipped offline word lens for a handful of pairs as a download. the experience by then was a feature inside a translation app rather than a standalone product.

quest visual closed as a company shortly after the acquisition closed. otavio good joined google as part of the deal. honestly, the standalone word lens app was the cleanest example of the acquisition pattern where the product dies into a feature inside the acquirer's larger product. for whatever reason, what people remember is the original viral demo loop. the long tail inside google translate that came after never matched the same brand-recognition moment.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »