wyle@inf.ethz.ch (Mitchell Wyle) (02/23/90)
> Any question on the quality of the picture is a question about the > quality of the witness. Don't you have to sign photos, tapes, videos, etc.? If it is doctored, the person who signed it goes to jail for purgery, no? It was always possible to air-brush or retouch photos. Computers don't make it that much easier. It was always possible to synthesize taped testimony. Since all this stuff needs a person behind it, computers won't hurt or hinder the current courtroom system. Mitchell
stuartw@tove.umd.edu (Stuart M. Weinstein) (02/23/90)
> I do not know how it is in the real world, but here in Wisconsin the law > on photographs is quite clear. A Photo is admitted as a part of the > evidence submitted by a witness. Does this mean that if the photograph was manipulated, the witness can be charged with perjury, as if s/he lied on an affidavit? Stuart
philip@pescadero.Stanford.Edu (Philip Machanick) (02/23/90)
I was doing a demo of DTP concepts a year ago, and happened to be using Image Studio. I hadn't used the program before, and was jut using it to illustrate a point, when someone in the audience said, "Can you make that tree disappear?" This I accomplished in about a minute. Imagine what a skilled person could do. Also, remember that the courtroom is but one context where this may be a problem. A corrupt TV station/politician etc. could do a great job of inventing "mud" for an opponent with a bit of creative editing. Don't tell me this isn't happening already. The issue is not the availability of tools, but checks on corruption inherent in a specific form of society (including the way courts operate), surely. Philip Machanick
bp@pixar.com (Bruce Perens) (02/23/90)
Walter Nirenberg writes: > .. investigating the impact of recent computer graphics technology > advances on the use of video and photographs as courtroom evidence. There was an apocryphal story going around the NYIT Computer Graphics Laboratory when I started working there in 1982. At that time we had a 24-bit paint system with NTSC scanners and high-resolution photo output, similar to what you can do on a PC today. It was said that a manager who wanted to void a traffic ticket had one of our painters alter a photograph to put a tree in front of a stop sign. The forgery wasn't good enough to fool the judge, who told the manager to pay up or go to jail. Now this could just be an entertaining story that someone there cooked up, especially since this particular manager was sort of unpopular with the research staff. I have no evidence that it ever happened. Bruce Perens ucbvax!pixar!bp
nazgul@alphalpha.com (Kee Hinckley) (02/23/90)
> So. What we need is information of a technical and intellectual > nature pertaining to this topic. Is there software available for > these inexpensive machines to enable average people to do this? > Has anyone seen examples of this kind of manipulation? etc. There is a very relevant issue of the "Whole Earth Review" you should track down, sorry I can't find my copy to tell you which, but it was sometime last year I believe (maybe?). Has flying saucers on the cover (faked of course). Kee Hinckley
gahooten@orion.arc.nasa.gov (Gregory A. Hooten) (02/24/90)
This type of deception has gone on for a long time, and there is a whole battery of tests that can be run to see if the photo is real. One is to check for the consistency of background lighting, or the abrupt changes in shading. Remember that to make a forgery, you had to scan it in in the first place, edit it, and then get it back to a photo. All of this can be done by an investigator. As closely as you can look at the photo, they can too. There will always be bits that were not changed to the correct pattern, or that were cut out of one photos background, and placed into another. These bits can be read. It is easy to fool many people, but it is more difficult to fool them for long. Greg
len@csd4.csd.uwm.edu (Leonard P Levine) (02/25/90)
Stuart M. Weinstein asked whether a manipulated or doctored photograph being admitted as evidence could result in the witness being charged with perjury, "as if s/he lied on an affidavit"? Sure does. Leonard P. Levine
buckland@cheddar.cc.ubc.ca (Tony Buckland) (02/25/90)
There's a thread in this discussion that says more or less "yeah, but you have to sign it and face possible perjury charges". This won't work for two classes of people; perpetrators who would face severe enough penalties if convicted that they have in their judgement nothing to lose; and the amoral types who would coldly calculate the odds of being caught and the rewards of getting away with it, and then cheerfully perjure themselves every day of the week if the numbers came out right. Tony
jsb@panix.UUCP (J. S. B'ach) (02/26/90)
Alan Sanders comments that "ImageStudio and Digital Darkroom can manipulate scanned or direct video input in a seamless fashion. And Adobe Systems has just released PhotoShop, which takes the capabilities of this type of software to new heights. Of course, none of these programs are intended to be used for fraudulent image manipulation. But I agree that the possibilities are scary." Alan, if by "fraudulent" you include ad agencies making products look better than they are, they most certainly are intended to be used that way. Interestingly enough, the CIA has shown interest in ColorStudio and ImageStudio, not, so they claim, because they would ever try to doctor photographs in this way, but because they want to be able to understand the doctoring process well enough to be able to detect the fraudulent images created by others. J.S.
jgk@osc.osc.COM (Joe Keane) (03/18/90)
Ludovic van B writes: > Maybe the only way to make sure a photograph is valid as courtroom evidence > is to ask for the negative. I feel this is - still - impossible, I mean > tampering with negatives or creating entirely false ones. If you can create a perfect fake photograph, complete all the artifacts you'd have in a real photograph, it should be easy to make a negative from that. I don't see how you could tell this from an authentic negative. Maybe there's some aging process in the negative which lets you date when it was exposed or developed? I don't know of any. > On the other hand, proving that a tape recording is 'modified' or entirely > false might be easier, I don't really know. Does anyone know whether computers > can say with total accuracy that a voice is artificial? (Imitated is a piece > of cake to detect) Even though the recording might be 'corrupted', i.e. > 'noisy'? I'd like to know. If you have a large amount of authentic recordings of some person's speech, you can make a recording of him saying anything you want. And if you're good, there would be no way to tell that he didn't actually say it. In many ways, this is even easier than making a false photograph. The only hardware you need is a PC with A->D and D->A converters. It might take a week of number crunching, but that's not a big problem. Joe