Digital detectives discern Photoshop fakery

New software combs for clues in al Qaeda tapes, Harry Potter pages, and celebrity waistlines.

By , Staff writer of The Christian Science Monitor

The magazine photo-spread showed French President Nicolas Sarkozy in swim trunks, canoeing across Lake Winnipesaukee, N.H. But something was missing.

Earlier this month, Paris Match ran this picture of the shirtless Mr. Sarkozy sans poignees d'amour – or love handles. With a slight digital nip-tuck, the magazine trimmed the flab that peeked above the presidential waistline.

"The position of the boat exaggerated this protuberance," explained Paris Match last week, after another French weekly, L'Express, exposed the touch up. "The correction was exaggerated during the printing process."

Recommended: Default

As image-manipulation software becomes easier to use and harder to detect, the problem of tampering has spread far beyond such celebrity "corrections." While fudged paparazzi moments do little more than embarrass editors, there are far more important – and sometimes illegal – fakes to catch.

"The most common examples of doctored photos occur in the media, but there are serious cases of image manipulations in security and investigations as well," says Cynthia Baron, author of the book "Photoshop Forensics," scheduled for release in December. "There are researchers working as the frontline of defense against digital fraud." And they're developing some very tricky ways to spot shams.

Over the past six years, computer science professor Hany Farid has become something of a digital detective. While Paris Match's virtual liposuction was exposed because the unaltered photo ran in several other publications that week (including the Aug. 6 Monitor), Mr. Farid doesn't need the original to reveal tampering.

As head of Dartmouth College's Image Science Group in Hanover, N.H., he's developed computer algorithms that can tease out the tiny flaws hidden in phony photos.

"There's no way to push a button and tell if it's real, but there are tests we can run that allow us to be pretty sure if it's a fake," says Farid.

Some of the investigative techniques are simply teaching a computer to spot what the untrained eye is too lazy to see. If a figure from one photo has been edited into another, there are almost always imperfections – subtle inconsistencies in the physics and geometry of the combined image. The vanishing points might be off, or the shadows cast from two or more objects may contradict one another.

"These are things humans are really bad at noticing," says Farid. But to a computer, the subtle differences are obvious.

Farid can now run possible forgeries through a gamut of tests, even checking the light reflections in people's eyes to triangulate the location of the flash camera that took the picture. If the analysis of subjects in a photo shows that the camera had to be in multiple places at once, the shot's a fake.

Courts face off with digital fraudsters

With the pervasiveness of computer editing software, investigators and courts are learning to deal with digital fraud. Since pictures and documents stand as the bedrock of evidence, Farid has applied his studies to help judges and juries determine what's real and what's been altered. Recently he testified in an intellectual property lawsuit. The plaintiff accused the defendant of stealing software and offered a computer screen shot of their programming as proof. After running tests on the evidence, Farid determined the screen shot was faked.

"They tried to fool the court," he says. "I think the case has now turned from a civil suit to a criminal case going the other way."

But after working on two dozen cases, Farid has found there are far more accusations of fraud than there are actual instances of foul play. In another case, a man insisted that someone had digitally added his signature to a scanned document. If the charges were true, then a computer could probably detect tiny discrepancies between the signature and the rest of the image. Farid could not detect any and concluded that the man had in fact signed the document.

How Al Qaeda alters its videos

Spotting inconsistencies in pictures is a major aspect for computer forensics.

"One helpful aspect of digital files is that they leave records, whether you know it or not," says Nasir Memon, a computer science professor at Polytechnic University in New York City. "Whatever you do to an image, it will leave tell-tale signs – artifacts hidden in pictures."

Scouring for these digital imperfections led computer security consultant Neal Krawetz to develop techniques of determining what specific areas of an image have been altered. He can even trace the history of those changes. Dr. Krawetz digitally dissected an image from a 2006 video of Al Qaeda No. 2 Ayman al-Zawahiri sitting in an office that's decorated by a banner with writing on it. (.) Through a computer analysis, Krawetz found that Dr. Zawahiri probably posed in front of a sheet and then was superimposed – much like the green-screen technology used in movies.

In fact, he concluded that the image is a composite of up to four different layers: Zawahiri, the subtitles and As-Sahab logo, the background, and finally the writing behind his head. "Apparently someone added those letters to the image afterward," he says.

How can he tell? If the picture was un­edited, the quality across the frame would be uniform. But many digital images lose precision every time they are saved. With each modification, older additions to the collage deteriorate. This disparity is often undetectable to the eye, but Krawetz's software can sniff out the variations.

The Harry Potter photo case

A few days before the release of the seventh Harry Potter novel in July, someone posted digital photos of every page in the final book on the Internet. The culprit has yet to be caught, but digital forensics experts know a lot about this leaker.

Embedded in each photo file is information called metadata, which reveals that the shots were taken by a Canon EOS Digital Rebel 300D. The tags disclose the year the camera was made, when the pictures were taken, even the serial number for the specific camera. "Now, if that person ever brings in his digital camera for repairs, they got him," says Bruce Schneier, CTO for the network security firm BT Counterpane. "This information is secretly hidden in all kind of electronics. Even [some Xerox] color printers hide information in printed pages that can be tracked back to your specific printer."

Mr. Schneier worries about the consequence such technology has on privacy. These buried codes are designed to track down crooks, but he warns that smart criminals will know how to scrub the metadata out of their files. The people they're hurting, he argues, are law-abiding citizens.

"Data is more and more traceable and now is being used to connect you to things that should be private," he says. "Do we really want ... cameras that link your pictures back to you?"

How to detect a fake

Spotting fake photos and forgeries doesn't always require high-powered algorithms. While image manipulation software is now easier to use, it still takes considerable skill to wield the tools well, says Cynthia Baron, author of "Photoshop Forensics."

"We're an intensively visual society, and yet we're not very good at visually scrutinizing," Ms. Baron says. But if you know what to look for, there are many ways to tell. Here are a few:

•Frankenstein images. A favorite trick is digitally sewing a celebrity's head onto another's body. Earlier this month, the Republican Party of Kentucky printed a campaign brochure with a fake photo of the Democratic candidate for governor, Steve Beshear, looking sleazy in a casino. The image is labeled "not an actual photo" and was designed to mock the candidate's stance on gambling. But the picture also demonstrates two clear signs of a stitched photo: an unnatural tilt of the head and an awkward seam where the head meets the collarbone.

•Too-straight lines. Few unmanipulated photos contain 90-degree angles. But computers love square corners. If a protester's sign has a perfectly straight edge all the way around it, it might be a fake.

•Recurring objects and patterns. After an Israeli airstrike in Lebanon last August, the Reuters news agency website ran a photo of smoke billowing from a building. But one of the thick puffs had a repeating pattern – an obvious signal, Baron says, that the photographer had copied and pasted parts of the image to make the damage look more severe.

•Truth in shadows. Manipulators often create collages without making sure that the lighting on all the objects matches up. An astute viewer can pick out such inconsistencies.

•If it looks too good to be true, it probably is. Andy Roddick surprised even himself when he saw the June-July cover of Men's Fitness magazine. In the image, the tennis star's biceps appear thicker than the Wimbledon trophy. "Pretty sure I'm not as fit as the Men's Fitness cover suggests," he jokes on his blog. "Little did I know I have 22-inch guns and a disappearing birth mark on my right arm." Oops.

Share this story:

We want to hear, did we miss an angle we should have covered? Should we come back to this topic? Or just give us a rating for this story. We want to hear from you.

Loading...

Loading...

Loading...