Like, say I wanted to search out the origin or further context for an image. Is there any protocol that would allow for copying and pasting an image itself (not text) and finding out more about it?
Is that even possible with today’s technology?
I’m sure there is a way to compare actual content but probably none that are close to being primetime image-search-comparison engines.
OTOH, it would be pretty easy to have a search engine that looked for the exact image, where you could upload or point to a picture and it would search through all its pictures to find the one with that exact checksum in whatever format you were putting in, if they just had a collection of images and their checksums in various formats.
There is an image searcher that’s being developed:
There was a Newsweek story that said someone used it to debunk a hoax. Apparently somebody photoshopped a giant snake into a picture of a river. The snake was some mythical beast that was supposedly 100 feet long and the photo looked convincing. But someone else used Tin Eye and found the original picture of the river, sans snake
I did once see an image search engine, but I can’t remember the name. It was pretty robust, too, being able to detect resizes, moderate color shifts, minor photoshops, etc.
EDIT: Yeah, Tin Eye was the name of the one I was thinking of, too.
There exist utilities that are supposedly capable of finding duplicate pictures on a computer, if would not take a lot to expand the concept across the internet.
I have not used any of those utilities, so I do not know how they do with pictures that have any kind of modifications from an original.(hope this is a forgivable hijack) Does anyone know/recommend any of these utilities?
Google image search used to have something like this. Below each thumbnail in the results (of a textual image search) there would be a link to “find similar images” or something like that. It would return images that were strikingly similar in terms of colours and shapes, but they would often be completely unrelated (eg. one picture is of a dog, one is of a building, etc.)
It was kind of interesting, but they seem to have removed that feature.
Image DeDuper.
I’ve used it, worked well. Wasn’t so great with a folder that contained a series of screen caps, thought they were all duplicates.
It’s free.
Another one that popped up searching for the DeDuper link, Image Deduplicator, also freeware.
CMC fnord!
Google image search does have a halfway kind of answer for this, you can search images by exact size, resolution, number of colors, image type (gif, jpg, etc.) It also has filters for clipart, photographs, news images, faces, line drawings, etc. you can specify quite a bit about the image you are looking for but it isn’t a complete solution.
Non-text search is a hot research topic right now, but as far as I know, there’s not much in production other than Tineye. I’d be interested to see anything else.
Song tapper is a clever example of non-text search.
TinEye looks amazing. Check out the Cool search function they have. Thanks for the great info!
It’s called Google Similar Images and it’s a program in beta testing. Perhaps if they had it before it still had kinks to work out and they are fine tuning the software. Cool, but not really what the OP is looking for.
It looks to me like that’s not actually image-based, though, but just comparing the keywords associated with the different images. Like, their first example isn’t looking for things that look like the Eiffel Tower, but for things that have the words “Eiffel Tower” associated with them.