While tinkering with personal photos to get rid of red eye is (usually) good clean harmless stuff, Scientific American has pointed out that fraudulently manipulating photos is a serious cyber crime issue. It’s so serious, in fact, that the relatively new field of digitial forensics was created to uncover evidence that separates the bogus from the genuine in the murky digital photo underworld.
Here are Scientific American’s top 5 ways that experts try and spot the fakes:
Lighting. While typically undetectable by the naked eye, a trained expert with the right equipment can determine if lighting conditions in the photo are consistent. In a real photo, light source directions would be consistent with reflections from objects. However, in a fake, there could be subtle – yet detectable – discrepancies.
Eyes and Positions. While eyes seem to be circular (at least in cartoons), they become elliptical as they turn to the side, up or down. As a result, it’s possible – not easy, but possible – for experts to trace the angle of the eye to see if an individual was in fact looking at the camera when the photo was taken. Experts point out, however, that it’s easier to use this method if an individual was “moved” from one side of a photo to the middle; it’s much harder when the move is slight (e.g. right side of the photo to slightly further right).
Specular Highlights. The next time you tell your beloved that his or her “eyes are like the stars,” you can – if you wish to totally ruin the mood – also mention that this is probably due to specular highlighting. And once you’ve apologized, you can even mention that you were referring to the phenomenon of how surrounding light in a photo reflects in the eyes in the form of small white dots. And after that – perhaps on the drive to the relationship counselor – you can mention that digital forensics experts check to see if the specular highlights are consistent with the lighting in the picture. When they aren’t, the photo has been doctored.
Cloning. Experts can hunt and (sometimes) find clones within a photo by using an algorithm that characterizes colors in a block of pixels. Clones are replicated images that, to the naked eye, look like distinct objects. For example, a digital photo of a large crowd may in fact have been a smaller crowd, but with ‘cloned’ pixels to make it look larger.
Fingerprints. Strangely, cameras have “fingerprints” that put a distinct signature on a digital photo. Experts can use a wildly complicated process (which we won’t even begin to describe here, because we can’t) to see if the camera matches the photo, and vice versa. Bear in mind, however, that the technique of manipulating photos by “changing fingerprints” is not always malicious. Scientific American points out that it can be done to improve the clarity of words on a page, on an ad, or something to that effect.
So while people may be tempted to doctor a photo (hopefully only for fun), remember that there are now methods to determine whether or not the photo is a fake.