CNET has a very interesting interview with Neal Krawetz, founder of computer security firm Hacker Factor. Of late Krawetz has turned his attention to digital images and the truths and lies they tell. This is particularly interesting to me in a journalism context.

Take USA Today. Every now and again, they put up pictures of Hillary Clinton and Barack Obama. And they will modify the pictures. I'm not sure who's modifying the pictures--whether it's the photographer submitting it or the intern who's putting them together or someone else at USA Today--but they'll modify it to increase the brightness, for example, on Hillary.

When you increase brightness on a picture, you bring out all the things like wrinkles that really aren't attractive. And they'll soften the picture on Barack Obama to make it look better.

And on helping publications avoid "modified" images.

In my talk, I actually give some pointers for the mass media like Reuters. If they really want to publish pictures that have been unmodified, here's how you can tell. One way is to use quantization table fingerprinting.

If the picture claims to be from a digital camera, and the quantization tables, which are used for compressing the image, don't match the camera, then you know that it's been manipulated. If Reuters had done that, it would have caught the fake photos.

Of course virtually every digital image, just like every darkroom developed image, is going to have some level of modification and manipulation. Burning, dodging, contrast, etc.. The question for me in terms of photojournalism is does it alter the fundamental truth and story of the photo. Photoshopping out a power line crosses that line between truth and fiction for me whereas upping the contrast a bit doesn't. I don't envy the photo editors that have to draw that line between truth and fiction but the lines do have to be drawn. As we've discussed before there are just too many instances of photographers in photojournalism situations taking huge leaps over that truth/fiction line to ignore the problem.