The Online Association of Photographers: People Are Really Bad at Spotting Fake Photos, Study Finds

Thursday, 20 July 2017

People Are Really Bad at Spotting Fake Photos, Study Finds

The Internet is filled with fake photographs, and some have even won contests run by the likes of Nikon. And here’s why: people are generally very bad at detecting when a photo has been faked.

A newly published study by researchers at the University of Warwick in England has found that people were only able to spot fake photos among real ones 60% of the time. What’s more, those who correctly identified fake photos could only point out what was faked 45% of the time.

The study involved 700 men and women being shown 10 photos — 5 of real and 5 faked in some way — and asked to label each one and determine where the faked ones were manipulated.

Here’s an example of a faked photo that was shown:

Can you figure out what isn’t real inside the shot? Answer: the boat in the background was digitally added:

Here’s an easier example to see. The right on the right is fake because the head has been flipped from the original (on left), putting the shadows on the incorrect side of the face:

Can you tell what’s wrong in this next fake photo?

Answer: the shadow was manipulated and not realistic. Here’s what the original untouched shot looks like:

If you’d like to take the same test as the participants to see how talented you are at spotting fakes, you can do so on the study’s website here.

“Subjects demonstrated a limited ability to detect original and manipulated images,” the researchers write. “People’s ability to detect manipulated images was positively correlated with the extent of disruption to the underlying structure of the pixels in the photo.”

“[…] our findings show, for the first time, that people have poor ability to identify whether a real-world image is original or has been manipulated,” they continue. “The results have implications for professionals working with digital images in legal, media, and other domains.”

The scientists are now working to see if people can be trained to detect fake photos better.

“The challenge now is to try and find ways to help people improve at this task”, Nightingale tells LiveScience. “We’re conducting new research to see whether people can make use of [telltale] signs to help identify forgeries.”