There’s good news for everyone who uses stock photos, today: finding the right one just got easier. That is, if you use Shutterstock. Shutterstock is taking a page from the Google Images play-book by implementing a reverse image search. Then, they took it a step further by giving us “visually similar” search.
Really, most of what you need to know is right in the names. Anyone who has dragged a photo into Google Images to find the original source knows what I’m talking about. Now, you can do the same on Shutterstock to see if they sell it.
The visually similar is where things get really cool. Found an image you like, but don’t want to copy the people already using it? Throw it up on Shutterstock to find images that are pretty close.
Both of these tools are built on machine learning concepts. The more people use them, the more accurate they will get. And make no mistake, it’s the images that are being analyzed, not tags or keywords, as before.
But obviously, I had to test it. Here’s what happened when I grabbed an image from Unsplash, and looked for similar images on Shutterstock:
Admittedly, that’s not an easy one. All the algorithm had to go on was the back of two heads, one human and one canine. But, if you look between all the smiling couples in the results, you’ll see a couple of shots of people with their dogs, including one from the back. Plus, nearly all the photos have that same warm summer feel.
This is what happened when I uploaded another photo from Unsplash. This one’s from NASA, so it looks a bit abstract, as you might expect.
They have the same photo in their library, and it’s the first result. The rest were basically all underwater shots. Not surprising, given the sheer amount of blue in the original. Look closely, and you’ll see some of the same color tones in the underwater life as in the land mass in the original photo.
Lastly, I gave it the black and white test. Because I couldn’t resist, I used a photo of my own:
This time, I have to say that the results were astounding.
Well, it’s new technology. The results you get may be hit-and-miss, based on both the algorithm, and what’s actually in the Shutterstock library. As previously stated, though, the technology should evolve over time, and get smarter.
Hopefully, this will lead to seeing a wider variety of imagery out there in the great World Wide Web.