• Quick note - the problem with Youtube videos not embedding on the forum appears to have been fixed, thanks to ZiprHead. If you do still see problems let me know.

Faking it

Southwind17

Philosopher
Joined
Sep 6, 2007
Messages
5,154
Given the almost universal progression from film photography to digital, leading to the impressive but fake images that one sees almost daily in social emails circulating the globe, mostly generated by some imaginative Photoshop afficionado, it got me wondering today: is it now generally much easier to convincingly fake a photo, and are digitally faked photos much more difficult to detect and analyze?
 
Given the almost universal progression from film photography to digital, leading to the impressive but fake images that one sees almost daily in social emails circulating the globe, mostly generated by some imaginative Photoshop afficionado, it got me wondering today: is it now generally much easier to convincingly fake a photo,

Yes. Used to take months.

and are digitally faked photos much more difficult to detect and analyze?

Not really for example spot the edit to this one:

http://upload.wikimedia.org/wikiped...of_the_Working_Class_-_Feb_1897_-_Altered.jpg
 
Given the almost universal progression from film photography to digital, leading to the impressive but fake images that one sees almost daily in social emails circulating the globe, mostly generated by some imaginative Photoshop afficionado, it got me wondering today: is it now generally much easier to convincingly fake a photo, and are digitally faked photos much more difficult to detect and analyze?

It's actually rather difficult to fake original digital photographs. The reason is that each pixel in a camera's CCD can actually only capture 1 color at a time. In order to make each pixel full color, the camera has to do some interpolation for each color, and those interpolations create characteristic correlations among nearby pixels. If one knows the camera that the photo was taken with (and that info can be stored within the file itself), then one can determine what these correlations should be, and test the photo against that. If you try to edit the original digital photo, you will probably destroy the correlations.

However... if the photo gets resized, color corrected, etc. then those correlations can get destroyed anyways. So it's not a simple matter to differentiate between an edited photo (ie, color balanced) and a faked photo.
 
It's actually rather difficult to fake original digital photographs. The reason is that each pixel in a camera's CCD can actually only capture 1 color at a time. In order to make each pixel full color, the camera has to do some interpolation for each color, and those interpolations create characteristic correlations among nearby pixels. If one knows the camera that the photo was taken with (and that info can be stored within the file itself), then one can determine what these correlations should be, and test the photo against that. If you try to edit the original digital photo, you will probably destroy the correlations.

However... if the photo gets resized, color corrected, etc. then those correlations can get destroyed anyways. So it's not a simple matter to differentiate between an edited photo (ie, color balanced) and a faked photo.
Sorry - to be clear, I mean digitally faked photos in their final printed media form, such as newsprint. That said, what you write above is very interesting - thanks.
 
Sorry - to be clear, I mean digitally faked photos in their final printed media form, such as newsprint. That said, what you write above is very interesting - thanks.

Which is why I mentioned that caveat about editing. Even online, photos almost always get edited, even if only to resize them (5 megapixels is now small for a camera, but it's way too large for a web page).

But it's the graphics software which really makes the fakery easy, not the digital photography itself, since one can do exactly the same stuff with scanned images from film. The scanning process might present enough of a barrier to stop a casual amusement-seeking faker who can't be bothered, but it presents no barrier to anyone who's intent on creating a fake.
 
This was the subject of a Scientific American article Digital Image Forensics

Barely a month goes by without some newly uncovered fraudulent image making it into the news. In February, for instance, an award-winning photograph depicting a herd of endangered Tibetan antelope apparently undisturbed by a new high-speed train racing nearby was uncovered to be a fake. The photograph had appeared in hundreds of newspapers in China after the controversial train line was opened with much patriotic fanfare in mid-2006. A few people had noticed oddities immediately, such as how some of the antelope were pregnant, but there were no young, as should have been the case at the time of year the train began running. Doubts finally became public when the picture was featured in the Beijing subway this year and other flaws came to light, such as a join line where two images had been stitched together. The photographer, Liu Weiqing, and his newspaper editor resigned; Chinese government news agencies apologized for distributing the image and promised to delete all of Liu¿s photographs from their databases.

Full article needs to be purchased, however experts can tell the difference. For example shadows that go in different directions is a give away.
 
I'm getting pretty good at detecting faked photographs. I can usually detect the fake ones on a professional site like photoshopdisasters.com.
 
I'm getting pretty good at detecting faked photographs. I can usually detect the fake ones on a professional site like photoshopdisasters.com.
Most people should be able to detect those, as the reason it's called "photoshop disasters" is that the errors are usually pretty darned huge; extra hands, missing legs, etc.
 

Back
Top Bottom