Secure Computing: Sec-C

Digital image manipulation is becoming a constant feature in the news. Recently Newsweek had an article about photo-editing services for your kid's school picture. Forget the retakes -- just pay $6 to have that grimace photoshopped into a sweet smile.

However, one photoshopped picture recently hit the mass media. A photographer in China reportedly took a picture of an antelope herd near a passing high-speed train. The Chinese government used the picture to show how the train was not having a negative impact on the environment. However, there is a problem: the picture was fake. According to the Wall Street Journal:

Earlier this week, Xinhua, China's state-run news agency, issued an unusual public apology for publishing a doctored photograph of Tibetan wildlife frolicking near a high-speed train.

The deception -- uncovered by Chinese Internet users who sniffed out a Photoshop scam in the award-winning picture -- has brought on a big debate about media ethics, China's troubled relationship with Tibet, and how pregnant antelope react to noise.

The WSJ pointed out a number of inconsistencies. However, I didn't see many of them. Here's what I noticed:

• Error level analysis (mouse over the image to see it) reveals four different areas in the image: (1) train and pillars, (2) a horizontal swath containing the antelope and ground, (3) a swath of ground above the antelope, and (4) the rocks closest to the camera. If this picture were real, then the entire picture should be near the same error level.
  
  
  
  (What about the sky? It is a uniform color, so I expect it to have an even error level.)
  
  
• Luminance gradient indicates three different light sources. (Mouse over the image and look for the dark-to-light transitions.) The first light source is on the train, tracks, and pillars. They all show sharp light changes along the edges. The second light source is on the antelope; they do not have the same sharp lighting as the train. The final light source is on the front rocks -- the light is hitting the rocks from the opposite direction compared to the train. Three light sources indicate three images spliced together.
  
  
  
  
• Antelope experts criticized the photo for having too many pregnant antelope with no young'uns. (E.g., the 2nd antelope on the left has a big belly.) They also said that the running herd was too tightly clustered.
  
  I'm not an antelope expert, but I noticed something else: all of the pregnant antelope seem to be running in single-file and have the same dark spot on their neck. In contrast, the males do not have the spot. While this could be specific to the breed, I think it may be the same animal. In fact, this appears to be a collage of frames. There were probably only 4-5 antelope photographed running by a camera that took a series of pictures. These individual photos were then merged to form one "herd" . Cynthia Baron commented that some of the legs may have been modified.

Following my talk at Black Hat, I was asked how many pictures were likely modified. Perhaps I am cynical, but I suspect that most of the popular, impressive, or significant pictures in the mass media are modified in one way or another. And soon, the "real" pictures of plain kids in elementary school year books will all be modified, too.