« Previous post
Next post »

December 19, 2006

Snap Image Search

Michael Arrington

35 comments »

Search engine Snap launched an image search feature today. The results are no better than that offered by the big guys, but Snap’s unique two column interface is something that some users find to be really cool. The problem with the lack of available photo meta data to assist with queries continues to plague all search engines, of course, leading to less than great results. It will be interesting to see if some of the new services launching, such as Polar Rose, can help by partnering with them and leveraging user generated descriptive data. More on Snap here.

  • Sphere It

This entry was posted on Tuesday, December 19th, 2006 at 10:00 am and is filed under Company & Product Profiles. You can follow any responses to this entry through the RSS 2.0 feed. Both comments and pings are currently closed.

Trackbacks/Pings (Trackback URL)

  1. TechCrunch Japanese アーカイブ » Snapで画像検索

    December 19th, 2006 at 1:34 pm

  2. 山丘突歐 » [新站] 強悍的圖片搜尋引擎

    December 20th, 2006 at 7:28 am

  3. Snap incorpora búsqueda de imágenes

    December 21st, 2006 at 1:32 am

  4. Tecnologia » Blog Archive » Snap incorpora búsqueda de imágenes

    December 21st, 2006 at 1:56 am

  5. Cloudberry Jam Fan Site Blog

    December 30th, 2006 at 7:59 am

  6. le jeu vidéo poker

    January 25th, 2008 at 9:15 am

Comments

RSS feed for comments on this post.

  1. Ian

    December 19th, 2006 at 10:11 am

    Nothing is particularly new with Snap but I do like their UI. It’s both distinct from the other guys and visually pleasing.

  2. Matt

    December 19th, 2006 at 10:15 am

    The effects are cool, but it seems a little much for an image search. It’s taking a long time to load for me.

  3. Greg Rumley

    December 19th, 2006 at 10:16 am

    Stunning!

  4. Abe Sultan

    December 19th, 2006 at 10:28 am

    This isn’t anything revolutionary. Live.com has had this interface since it launched months ago

  5. shortshire

    December 19th, 2006 at 10:37 am

    This is really nice to use. I think it’ll give Picasa a run for it’s money for user interface. the only part that might hurt is the load times of the pictures but we’ll have to wait and see.

  6. Gary M

    December 19th, 2006 at 10:52 am

    Usability is excellent — the integration of the mouse wheel for scanning through photos quickly (even on secondary pages) is excellent. I use google image search a lot, I’ll switch to this and see how it goes overall.

  7. Jon

    December 19th, 2006 at 11:03 am

    That’s pretty good, but I’ll stick with a meta-search…this site has a good image one which you can also customize: http://www.srchr.com/?PageID=7.....0Arrington

  8. JHSA

    December 19th, 2006 at 11:09 am

    image recognition is one of the most difficult tasks … even a “simple” think like OCR is dificult to implement, so it would probably would take a while for this to work perfectly, or within reasonable error margin, if ever.

  9. John

    December 19th, 2006 at 11:16 am

    Its really simple and convenient to use. You don’t have to click on the image to view a larger image. Search image in Google, the image are small, in order for you to view the image - you would have to click on twice. For Snap Image search, all you have to do is click on it once, the image is enlarge and clear - in the right panel (talk about slick, eh!)…

    Yes, I do agreed that it is slow when loading, but once the images are loaded, its stunning.

    John

  10. Patricia

    December 19th, 2006 at 11:34 am

    @JHSA, it really is. It’s going to be fun to see how the technology advances now that it’s been brought to market.

  11. Libby

    December 19th, 2006 at 11:54 am

    Interesting. Looks like a web version of Beholder for the Mac:
    http://www.macupdate.com/info.php/id/13248

  12. Frank Cefalu

    December 19th, 2006 at 12:41 pm

    Why cant companies incorporate better Meta Data Filters. It isn’t that difficult to string the keywords of a Photograph.

  13. Stephen Kellett

    December 19th, 2006 at 1:44 pm

    But if meta data for images was available, what would happen?
    Very quickly the spam merchants would most likely pollute the space with their poison. Nice images with completely inappropriate meta data.

    I think we’ll be better off waiting for Google and various startups to solve
    the image recognition/classification problem.

    (BTW. Michael, I emailed you about a month ago. Never got a reply. Did you get the email?).

  14. Michael Arrington

    December 19th, 2006 at 1:58 pm

    Stephen, please re-email me. or ping me on skype.

  15. Fashion Industry Ceo

    December 19th, 2006 at 2:19 pm

    it seems all image search engines..pretty much suck…from yahoo-google

  16. basicity

    December 19th, 2006 at 2:51 pm

    Wow, what an awesome 4-letter domain name and very relevant to what they do!

  17. Dan

    December 19th, 2006 at 3:04 pm

    I like the 2 part screen but I just searched for images of 2 different musicians that I like. In each case I keep getting popups of an “authentification required” screen (not the same one in each case). This is very annoying. It would be nice if Snap could stop this.

    Type in James Taylor and click next for an example.

  18. Jared

    December 19th, 2006 at 3:16 pm

    As I understand it, Google and Yahoo would love to adopt this interface, but have not done so because of copyright concerns. There is no question that the interface of Google and Yahoo images is annoying to users, requiring an extra click and a costly page load.

    Perhaps someone with legal experience can elucidate how cases like Google v. Perfect10 have shaped these interface decisions.

  19. artbyvanee

    December 19th, 2006 at 4:15 pm

    I did a search on the one thing that I’m the world’s best expert on: me. Also, I have a serious art hobby with a substantial art website with distinct colorful art at http://www.jonathanvanee.com (check it out!). Snap didn’t pick it up at first, but thanked me for my search. After a little while, a preview of my search appeared, along with three links. All links were relevant - which is unlike google and yahoo, where you get a huge amount of garbage along with your search.

    Hey Michael, I’m a fellow Wilson Sonsini alumn who also has a substantial web presence (obviously not as big as yours though). I’ve sent you an e mail to the editor@techcrunch.com address and would love to hear from you.

  20. Adrian Keys

    December 19th, 2006 at 4:30 pm

    Cool ….really cool. Something about the site sets it apart from other engines…I much like the concept of the 2 panels and how you are able to pick a topic and scroll right through the images…really neat.

    http://www.jollyjo.org

  21. Mike

    December 19th, 2006 at 6:55 pm

    Though I believe they have put much efforts to the user experiences, it’s still a little bit complex for a searching, compared with big search engines. For webmasters, the preview of the link service is very cool, and I think many webmasters/bloggers will add it soon, just as Techcrunch did.

    Tech Tutorials: http://www.hotcoding.com

  22. The Alpha Project

    December 19th, 2006 at 8:32 pm

    The power in image search (and video search for that matter) will really come down to integration with the other services that will use them. either for a business purpose or for social networking. If they can get on the ball with some great ways to integrate with other services then it would go a lot further.

  23. SearcH Engines WeB

    December 20th, 2006 at 5:07 am

    http://search.msn.com/images/results.aspx?q=flower

    Compare the ASK layout and results to the MSN LIVE version of Image Search

    …Interesting

  24. Markus

    December 20th, 2006 at 6:35 am

    Great. Except the reason no else is doing this is because it is a blatant copyright violation, not to mention dishonest to the website where the image was found (since they get no traffic, yet still pay for the image bandwidth).

    On top of that, since Snap pre-loads all the images, that means third party sites get dinged for bandwidth even if their image never gets viewed!

    This is outright bandwidth theft coupled with countless copyright violations.

    If you are a webmaster who has ever dealt with annoying people hot-linking your images, you know how much this sucks. Lame.

  25. Xenia

    December 20th, 2006 at 7:18 am

    Google has a different approach to get image meta data. It is not very well known, still in beta, but great fun, have a look at

    http://images.google.com/imagelabeler/

    Two users (unknown to each other) try to give best labels to images. If they match, the get the next picture. Try it, it is really fun.

  26. Martin Seebach

    December 20th, 2006 at 9:24 am

    Image searching is done on meta data not images? Seems to me much better results would come back using the pictures. Something along the line utilized in fingerprint comparison. Prolly one more thing that shows how un-plugged into the ways stuff works I am.

  27. Frank Cefalu

    December 20th, 2006 at 11:38 am

    @ Stephen

    There will always be spamming. Whats wrong with looking at the meta data of images. If anything it will reveal data not usually found due to faulty alt tags, and image names.

    Sure, if spammers want to waste time modifying every photographs meta data to get on the search engines image search they could, let them . Spamming isn’t anything new. But right now that problem is non-existent. When it becomes existent than you deal with it accordingly.

    No point in dismissing a photographs meta data for fear of a problem that isn’t present yet.

  28. Alexei

    December 20th, 2006 at 5:39 pm

    Re: missing metadata and image recognition

    This search engine does have a nice UI!

    As for image recognition, the current techniques are still in the ‘blue sky research’ zone. However in some situations they may already be useful. We have been working for a few months on a prototype web image search engine called Behold which assigns likely keywords to images based entirely on image content. This is done using statistical techniques. When this is combined with regular metadata search results often improve. There are some example queries on the website.

  29. David Mackey

    December 23rd, 2006 at 3:15 pm

    They could just pay a few of us to meta tag data. :-) I used to do that for websites for StudyWeb (now LightSpan).

« Previous post
Next post »