Images get posted (er, linked to) from time to time in GQ to question their validity. Fakes are detected by looking at shadow lines, warble effects, dipthinter scalage, and who knows what else I could make up. My ignorance aside, posters point to telltale signs of photo manipulation via Photoshop or whatnot. But has the art and science of digital manipulation grown to the point where an expert can’t tell the difference? That is, given modern equipment, software, time, and training, could a person(s) create a significantly doctored image such that someone else with similar resources couldn’t detect the change? I’m thinking more than removing redeye—more like completely removing or inserting a person in a scene or something like that.
If things haven’t progressed that far, what are the technical problems? Are they surmountable? Will they ever become automated and simple (i.e., a lay person can undetectedly change a photo)? Will that have any bearing on photographic evidence?
Thanks,
Rhythm
Yes.
It’s just manipulation of pixels, and if you put the right pixels in the right place, it can look like anything you want it to, as photoreal as you like.
The “telltale signs” is just inferior skill and talent at achieving the results. Most people make those mistakes some of the time, but plenty of people can and do make the effort to achieve a seamless indistinguishable result.
There is research being done into the underlying noise signature of a digital image, which promises to make fake detection much easier: http://cosmiclog.msnbc.msn.com/archive/2007/02/19/60203.aspx
As the article says, though, only until the fakers understand the method, and start adding the right kind of noise to the fake picture.
Because it’s digital info, with enough time and effort, any kind of fake-detection can be overcome.
That’s not to say fake detection is useless-- the more tests a faker has to defeat, the more chance they’ll screw up somewhere-- but no test will every be a guarantee.
Why yes, I believe it probably has.
Look here. Some really interesting (and bizzare) work.
Or, more likely, insufficient effort. Even if the faker does have the skill and talent to produce a truly seamless image, who’s the audience? If you’re trying to put together something that’ll fool the FBI to frame your archnemesis for a crime you comitted, then you’re going to want to put in quite a lot of effort indeed. But if you’re just trying to fool some credible saps on the Internet, why bother? And if you’re just editing the guy who was absent that day into the group photo you’re going to hang up in the hall, it doesn’t even matter if there are tells obvious to the naked eye.