If you haven't heard by now, newspapers and blogs are reporting that a photo of an Iranian missile test yesterday was digitally manipulated. There were allegedly just three missiles in the original photo, but four in the doctored one.
The New York Times reports that Agence France-Presse obtained the image from the Iranian Revolutionary Guard's news Web site, Sepah News. The Associated Press later released an identical-looking photo, except instead of a fourth missile, there was a military vehicle on the ground with a missile on its back ready to fire.
Hany Farid, a digital forensics expert at Dartmouth College, wrote an article about faked photos for our June issue. We called him for his take on the missile photo and some other images from a tornado video that some say are forged.
Here's an edited transcript of what Farid had to say.
Do you think the picture with the four missiles has been altered?
It's pretty clear that all four missiles didn't launch at the same time. The question is whether this is a straight clone job [copy and paste], like The New York Times blog is suggesting, or if the fourth missile is in fact a separate missile launch that was photographed and then composited into the original picture.
What are some of things that tip you off in these two pictures?
Well, at first glance it looks like the second missile from the left and its trail was possibly copied in the original and then pasted in as the second missile from the right in the edited version. But I'm not so sure it's that simple a story. If we look at the trails from these two missiles, for starters, there's a black dot just under the second-from-right missile that's not there on the other one.
Then there's the smoke plumes rising from the ground. If you look at the smoke plumes underneath two rockets on the right, those folds of smoke on the right-hand side of the trail look pretty similar, too, though. But if you look very closely, they are not identical; the pixels don't line up exactly. This distortion could have happened when the JPEG file [a common kind of digital image] was compressed, so it could just be cloned image. Or, it may be that the same kind of missile can make a very similar-looking plume.
Something else to notice here is that in the edited version, the rockets look a little bigger and thicker. This means that they are closer to the camera or the ground. So, it's a possibility that that so-called edited and the original are actually different shots entirely, taken by two people who took the pictures almost at the same time but from slightly different distances from the launch.
So what is the verdict on the picture with the four missiles in it?
It's almost certainly doctored in some way, but there are some subtleties here. Whoever did it did a reasonably good job.
What about images from the "faked" tornado video?
It's sort of the same story with the missile images; they are close to one another but are not the same copies. What's being claimed as the original video here from four years ago—the one with the orange sky—it's actually very blurry. The alleged fake, though, appears to be a higher-resolution version. This is not what you expect; if you add information to a picture or change it, it tends to lose resolution.
Secondly, if you overlay the two images, which I just did in Photoshop, there are definite similarities in the tornado's kinks, but they are not exact. So it's not like someone just flipped the image around and settled for that. This raises the question of how variable are the paths of tornadoes? In other words, what is the likelihood of two tornadoes looking very similar to each other? It doesn't help that the overall quality of these images is grainy, Loch Ness monster–style material.
What is the verdict on the twister pictures?
The jury is still out on this one, but there's definitely enough evidence to be asking questions. Then again, there are coincidences out there. It wouldn't be the first time that two pictures look about the same.