How to find all/any copies of an image on the web?
October 29, 2007 2:58 PM   Subscribe

I have a jpg file. What is the best way to find any instances of this file posted anywhere on the web. This is pretty much a dupe of this AskMe from 2004 but i was hoping technology might have advanced enough to provide a new answer.

I realize it would be very difficult to find resized versions of the file. But finding exact copies, albeit with differing file names, seems technologically feasible.

Surely it would be easy for google/yahoo/whoever to calculate some hash of each image in the database, which would then allow me to calculate the same hash on my file, then search by that hash? Does anyone know of a site that allows this? I am a programmer, so would not shy away from using an API rather than a nice gui interface.

Failing that, I don't understand why no one will let me search for images by size. And I mean exact pixel size, not the vague (small, medium, large and extra large) categories that all the search sites have now.

The answer to 'why would i want to do this anyway' is that someone has posted a nsfw image of a friend of mine to the web. I don't know the original place it was posted, but I do have a copy that someone reposted to a forum. I would like to search out any other copies, and would be even happier if in doing so I find out who originally released it into the internets against her wishes. The reposter has taken the pic down and 'thinks' they got it from imagevenue or some site like that, so any solutions specific to imagevenue would also be potentially helpful.

As always, thanks to everyone who answers.
posted by mrgoldenbrown to Computers & Internet (12 answers total) 2 users marked this as a favorite
Google Search By Filetype.

For example, if the image name was image2384.jpg, you would do a search for "image2384.jpg filetype:jpg" (minus the quotes).

Of course, if they change the name of the image, this won't work in the least. And for pages that are not indexed by Google, this won't work either.
posted by JPowers at 3:06 PM on October 29, 2007

Actually, now that I think of it, the search command would instead be "image2384 filetype:jpg" (minus the quotes). You don't need the .jpg at the end of the search phrase as I had above. Try both to be safe.
posted by JPowers at 3:10 PM on October 29, 2007

JPowers, finding files of the same filename is not what Mr Golden Brown is wanting to do.

Mr. Golden Brown wants to find the image file, even if the filename has been changed.

Mr. Golden Brown: But finding exact copies, albeit with differing file names, seems technologically feasible.
posted by jayder at 4:46 PM on October 29, 2007

jayder is right. i've already googled for the filename i have, which I know is not the original.
posted by mrgoldenbrown at 4:57 PM on October 29, 2007

I have a friend who works for a visual seach company called Idee. They have a service that might accomplish what you need, though I imagine it is not free. They are geared towards content owners like stock photo companies and such. However, it can't hurt to write them an e-mail.

In fact, I will mail my friend and see if there is anything they can do for you. I'll mail your profile when I hear back.
posted by PercussivePaul at 5:05 PM on October 29, 2007 [1 favorite] might be able to help you out. I have never used them yet but I hear they can do some pretty good stuff.
posted by bkeene12 at 7:23 PM on October 29, 2007

I think the reason why google et al don't already have hashes of pictures, is because computing hashes is expensive.
posted by philomathoholic at 11:42 PM on October 29, 2007

is there a way to find duplicates on your own pc (with different file names)?
posted by freddymetz at 3:03 AM on October 30, 2007

To solve this problem on my own pc, I would start by doing a search for all jpg's of XXXXXXX bytes or lower. I could then sort the results by file size, and look for an exact match by examining the thumbnails. That is the quick and dirty way, using only XP's built in search feature.
posted by mrgoldenbrown at 9:44 AM on October 30, 2007 [1 favorite]

My friend says "Yes, we could likely find other copies of the image out in the intertoobz if they exist. But our image monitoring service is typically used by companies or individuals wanting to monitor thousands or millions of images--and our searches are most efficient that way. Single image searching throughout the internet is currently quite costly to us, so we don't offer it on a wide scale yet."

However she is going to inquire and report back to me if anything comes up. I will update if I hear back.
posted by PercussivePaul at 12:28 PM on October 30, 2007 [1 favorite]

If you think that it is in imagevenue, you will want to construct a google search that will look for stuff on that site.

Also after visiting that site and looking at how images are returned ?image= in the url you can use inurl:?image=

to if not find all images at least narrow it down.

I believe using both of these will only search the web for locations of the images in blogs that it crawls. if you head over to google though and search for [site: inurl:?image=] everything in the [ ]'s without the [ ]'s

You will come up with some interesting albeit seems to mostly be nsfw images, not all that shocking, but hopefully one of these girls will be your friend. Or maybe not because now everyone on MetaFilter has her picture ;)

Direct link to googlesearch

Also this technique can be used on any site, if you feel you know where it is hosted, you can hopefully find it this way.
posted by Joe.At.Trends at 3:01 PM on October 30, 2007 [1 favorite]

I almost forgot to mention, I assumed that the imagevenue was way too underused to have been around on the interwebs for even an hour. That search above will return around 5 or 6 pages, I believe even with the safe search on. While you search the web for this pic, you will need to turn off safe search.

Those images are only images hosted on sites as links. not shown as an image. This would explain the low return rate. If you follow that link to google using that same search string click on image search at the top. This should search for all images that are shown on pages, and give you the page they are shown at. Also remember to turn off moderate safe search, make sure you dont' filter any searches at all.

This should provide you with a way to hunt down the image, given lots of searching however, you will have to look at all the thumbnails shown. But this will provide you with the forum/blog that it was posted on, you should also be able to see who posted it, and find out if that is a first generation post or maybe get some more information from that person on where they found the image.

Keep in mind anyone who preforms this search it is for a NSFW image of his friend so don't open these at work. (Or even in front of your significant other if she is like mine)

here is the direct link to google images. Good luck and I hope this helps.
(this link includes the safesearch off portion, proceed with caution)
posted by Joe.At.Trends at 3:38 PM on October 30, 2007

« Older Short world-changing documents written in English.   |   How to scale corner radius based on a logarithmic... Newer »
This thread is closed to new comments.