Xsights is an Israeli startup that develops interesting technology that allows for transformation of static objects like billboards, posters, printed images and more into interactive multimedia experiences using your cellphone camera. Theoretically, you could use Xsights to dynamically obtain more information based on e.g. words from a newspaper article, a landmark, a photo of an actor on a movie poster, a street sign, etc.
This bears some resemblance to augmented reality applications for mobile phones, except that it’s not really real-time, since camera shots are analyzed in the background after saving photos to the phone.
By pointing the phone camera to objects and saving shots to the program, users should be able to gain information such as directions, prices, videos clips, audio files but also purchasing opportunities, digital coupons, and more.
At least, the technology seems interesting on paper. In reality, there’s a big caveat: all the associated content needs to be preset by third parties on beforehand. That means it currently doesn’t recognize stuff I tried to test the app, such as company logos (Sony), pictures of celebrities (Jennifer Connelly, Brad Pitt) and a couple of movie posters (easily identifiable ones like the ones for Batman and The Godfather). All I got was a cryptic error message because no one has associated the images with anything so far.
It’s a far cry from what the demo and this video makes the mobile application look capable of. Xsights is looking to set up B2B partnerships with publishers, other content providers and advertisers to expand their database of recognizable items.
Xsights is today also launching a user generated campaign for individual users, basically giving them the opportunity to add creative touches to their interactive experience by “attaching” video clips to invitations, greeting cards, photos or videos – making the associated interactive content personal.
Xsights is giving 1,000 TechCrunch readers exclusive invites to be the first users to upload to the UGC Campaign. Just head on over here; only TechCrunch readers will be able to upload at this point.
Additionally, users will now be able to watch trailers for new movies currently showing in local U.S. theaters by capturing an image of a movie poster with their camera phone. In contrast to my experience with random images, analyzing the posters listed here worked just fine.
Xsights supports 3 ways of connections: Video Call (clientless), iPhone, BlackBerry, and selected J2ME handsets, and also works with MMS on all mobile handsets equipped with a camera.
Try it out and let us know what you think.
Great. too bad i don’t have an iPhone
guess i’ll have to wait for the 3G call and have the market embrace it here in Israel..
did you know you can d/l the xsights straight from the iTunes app store?
http://appsto.re/xsightslight
this site is fun too: http://appsto.re
There have been tens of companies trying this kind of b2b mobile experience with different and even simplier interaction from the user (shortcode dial, barcodes) but he challenge behind building enough “advertisers” before you have a single mobile customers and bring enough users before you have a single advertiser killed all of them. Why it will be any different?
Hi Eyal,
First off, you are right that there’s a challenge of getting to market acceptance (I guess that this is true for any new technology). Naturally, we can’t be sure that xsights (and other visual search players for that matter) will succeed better than barcodes.
For one note that barcodes do well in Japan and other parts of the world. They are also very useful for one-off usages (like plane e-tickets, coupons etc.) where the “end-user” is a machine
The current offering we (and others in this space) have is that of “visual tagging”. i.e. you have find a database of images (or manually add them) where the links are pre-defined. xsights already offers an advantage over barcodes and similar methods by not requiring adding or altering the activated “thing” wheres you have to embed barcodes in the image. There’s a caveat here that we need to activate complete databases (e.g. all the currently playing movies) so that you’d know you can take an image to view -otherwise it is more similar to barcodes, even though it is more visually appealing.
xsights, has a wider vision than that for which barcodes and similar methods will never do – We also look to enable visual search – where you don’t depend on the cooperation of others to grow your database and you can crawl the web and index images just as search engines do for textual data. Yes, we are still far from that vision but we have begun some work in this area and already have a couple of related patents (pending)). You do have to start somewhere and we think there’s enough value in visual tagging both in the b2b space and the b2c space.
You can also see the visual search advantage by the user generated capabilities we launched today which lets anyone activate and tag any image (ok, ok almost any image
) something you won’t be able to do with barcodes and the like.
Arnon Rotem-Gal-Oz
VP R&D, xsights
Cool technology, but must have more content ASAP
Bears as opposed to bares (unless the resemblance is easy on the eye that is…)
Wow, thats like the coolest thing since sliced bread!
Jess
http://www.anonymous.ua.tc
It needs lot more work. All the knowledge needs to be made into neural network with image recognition capabilities.
For example.
This photo is of Jenniffer Connelly. (Image recognition that can be input only by a human; computer doesnt do names).
BUT
The date of birth of Jenniffer Connelly is 14-JUL-1976. (This is a factoid that can be input through a collective human effort).
Now when you look at a picture of Jennifer Connelly and wonder what her date of birth is, you swipe your mobile camera across the pic and ask your question to an intelligent bot which looks up the factoid like answers.com (natural language query).
The same thing can be done with mouse gestures and baloon type opening tooltips. Say you come across the words “Egg muffin” as part of a text in a web page. Then you suddenly get the urge to visualize how an egg muffin looks like (If you havent seen it before; like a kid). Then you would select the words “Egg muffin” and use a finger gesture in your mouse’s biometric touch pad. In this case, you would do the finger gesture “V” (for visualize) in a touch pad located as part of the mouse. Then the image for egg muffin appears on screen. Similar commands- M = Map, FS = Find Similar items using Google Sets. For example. “Four Seasons” -> FS command would bring up Hilton, Ritz Carlton, and other hotel chain names.
I agree, there’s still a lot more work to do – but you’ve got to start somewhere…
We thought that letting everyone link whatever they like already gives some value and we’re working on increasing the value of the application (e.g. by adding more content types)
Arnon Rotem-Gal-Oz
VP R&D, xsights