In other words, considering how widespread and easily accessible photo manipulation technology has become, how can a professionally doctored photograph be distinguished from a legitimate one? And if there is no foolproof method, how then can any photograph be submitted as evidence of anything?
Most criminals wouldn’t have access to the sorts of resources where it’s relevant. And the police are too busy to put any effort into manufacturing evidence. They only need to take the evidence available, hand it over, and let the courts deal with it.
My understanding is that it is possible to tell (at a pixel level) whether a photograph has been doctored.
And, Sage Rat, needn’t be a "criminal’ who doctors photos. “Here’s a picture of my wife giving a large amount of cash to a known drug dealer, so I should get custody of the kids in the divorce settlement, your honour.”
Think of it this way: it can be photographic evidence without being photographic proof. And even if photographs can be doctored, some can be doctored more easily than others.
There are cameras made with verification data added to the picture when its taken to check nothing was altered afterwards - a device compares the picture to the data to check nothing has been changed. While nothing is tamperproof, they would require more effort than a normal picture, and used for evidence gathering etc. Also video will generally require more effort to doctor as instead of one frame you have to change many, without any errors apparent.
Its worth remembering that its also pretty easy to doctor a witness statement or any other number of evidence items. Thats why generally we rely on multiple pieces of evidence building a picture rather than single items.
Otara
Photo evidence is one of the most common pieces of evidence. Anything can be questioned, it is simply something courts deal with. It is far less likely to be messed with than witness testimony.
Wouldn’t that be perjury, and therefore a crime?
As for the OP, trick photography is almost as old as photography. Photographic evidence can be entered, it can be disputed. A party in a legal case, criminal or civil, can say that the photographs are fakes. And there are experts at detecting fake photographs that can be called to give evidence to the reality or fakeness of the evidence.
Pretty much any kind of evidence can be faked. Should we reject all evidence because it *might *be faked?
If you’re interested in the topic, I’d recommend a virtual stroll to the Image Science Group at the Dartmouth College. Their website has links to a couple of published papers, like Digital Doctoring: can we trust photographs? that deal with the matter in detail.
Photographs are probative or dis-probative of a fact in issue. If they are doctored that is a matter for cross examination or rebuttal. Witness Testimony is routinely shown to be defective due to inaccurate recollection or a misinterpretation of circumstances or outright lyining. The deficiency do not prevent witness testimony from being considered, it is for the cross examiner to point these out and impeach their credibility.
I’ve mentioned this on the board before, but: photographic evidence isn’t introduced without testimony backing it up. That is, you can’t just toss some photos to the clerk and say, “Admit these as evidence.”
There are questions that have to be asked of witnesses before physical evidence is introduced. We call these questions “predicates.” The predicates for photographic evidence are among the shortest; basically, you have to ask, “Do you recognize what I’ve just handed you?”, “What is it?”, and the deal-sealer: “Does that photograph fairly and accurately depict what you observed on Date X?” After that proper predicate has been established, the photographs are reliable enough to be introduced as evidence, though the other side may try to question their accuracy if they wish.
Interesting stuff, though I suppose this was better suited for GQ than here.
wintertime’s links are especially informative. Thanks for the responses.
Someone mentioned about verification data with photographs, which is something that’s starting to become common, particularly in areas of surveilance where the likelihood of images being used as evidence is more likely and, thus, securing their authenticity is important.
But there’s more to it than that. Digitally modifying images leaves artifacts. These may not always be visible to the naked eye, but images are essentially nothing more than huge matrices of numbers and manipulating them changes various trends like derivatives, edges, etc that are trivially detected through a little digital analysis. In fact, in various Computer Vision courses I’ve taken, we’ve spent a fair amount of time specifically covering those sorts of topics.
The other thing to consider is, how are images being manipulated? It’s one thing if you’re touching up an image of a model to make her a little thinner or remove wrinkles. That software is specifically designed to take into account those sorts of artifacts so that it can make those changes without significantly distorting the image, but how likely is that sort of image manipulation going to come up as evidence?
OTOH, you might see something where someone photoshops an image and, perhaps, puts a gun in someone’s hand or whatever, but that, in many cases, will leave image artifacts that are extremely difficult, if not impossible, to cover up unless done by an expert. You’re not just manipulating existing trends in a photo, but you’re inserting or removing them entirely, which will almost certainly leave detectable artifacts. For instance, if the lighting is not from exactly the same angle, it can probably be detected, if not somehow seeming visually off. Even small differences in color saturations, edges, sharpness, and all that sort of stuff is detectable.
In short, even with modern technology, it’s a lot more difficult to fake an image than you might think. Sure, we’re getting better at making authentic looking fakes, but along with that comes the technology to detect these fakes.
And, of course, beyond that, as others mentioned, images are part of a large case, usually including multiple images and, of course, testimony that identifies them. Sure, if you’re trying to convict someone on a single fuzzy frame from a security camera without any real testimony, it’s difficult to detect if it was manipulated–hell, the whole 9-11 Pentagon thing is more or less assuming that–but if that’s the whole case, well, it wasn’t very strong to begin with.
I would be more concerned about the FBI (or another agency with deep pockets) manipulating evidence for corrupt purposes.