Product: Camera: Digital: Annotation
DLSR Photoshop-evident Photos   (+6, -1)  [vote for, against]
To ensure you know if it has been digitally altered.

This is a Digital camera, in which the photo has been digitally signed, and watermarked, within the camera itself.

It should also have a GPS receiver for time stamping, and location of shoot.

in addition, it has a 360 degree lower rez camera mounted on the top of the DLSR, to capture the context of the photo. (or perhaps just a wider view fish lenses)

How this works is that the DSLR's sensor has two raw image line out to two different processors, a normal output for non-protected images, and a encrypted link to a crypto image processor. With this crypto image processor, there is less an emphasis on image quality (e.g. resizing it for jpg), and more emphasis on ensuring the photo cannot be easily tampered with. This is by signing the photo, and watermarking it with a serial code embedded within this special processor.

(some of the information you may want to encrypt into the photos, are the name of the photographer, location, time/date, and how it is taken.)

The reason you want this, is to ensure the integrity of the source is not compromised, for mission critical applications. E.g. Crime scene investigation. Or even a photo competition, where you need to screen thousand of photos against cheaters. And last but not least, war photographers, and investigative journalist.

We need to do this, because these days, people are ever more willing to distort the photograph to advance their own agenda, e.g. the cropping of the photos, during the "children overboard" controversy in Australia.

Just because its protected against tampering, doesn't mean it doesn't allow for some minor approved modifications. In fact it does allow for modification, however it embeds the original photos in the meta-data of the modified photo, and specifies what modification has been done to it.
-- mofosyne, Apr 02 2010

this post looks shopped.
-- shapu, Apr 02 2010


"Cropping" is kind of special because it can be done in the field - by selectively photographing only part of the picture or, in a more general sense, by only going to see one side of a conflict. So, there's always some artistic manipulation.

For forensic purposes, this seems like a good idea. The market is kind of small, but I wouldn't be surprised to find something like this for real.
-- jutta, Apr 02 2010


When I worked for Kodak systems like this were discussed, especially in the new business unit I worked for.
-- Aristotle, Apr 02 2010


Ah, then how about this! If this standard become 'popularized', then maybe a browser extension can be released that automatically checks the signature of the photo and authenticate the image. E.g. Hover over an image from a news website, and it will process it for a few seconds before showing you its 'certificate of authenticity'.

It will help ordinary citizens to better trust photos taken by newspaper online, just as we have "https:" to show that a 'banking website' link is secure.
-- mofosyne, Apr 03 2010


Photojournalism doesn't get much photoshop - it's the selection of the photo editor working with the editor that misleads/sets the agenda/illustrates the point of view. Glossies on the other hand...
-- wagster, Apr 03 2010


[+] for you and another [+] for Akimbomidget.
-- Postscript, Apr 05 2010


would it help if the standard also calls for the camera to have another low resolution 270 degree camera, to allow people to know the context of the photo? (no frontal shot, since that is already taken by the photographer main lenses)

basically, it also ensures to a certain extent, prevent photographers from 'staging' a shot. (or at least, force them to acknowledgment it'.)
-- mofosyne, Jun 30 2010


a small mirror ball, positioned on top of a lower res camera? (with parts of the view restricted to 270 degree)

It will embed a polar image into the photograph. Using the right software back in the computer, you can flatten the polar image.

(now that I think about it, its prolly more cheaper to just go for a 360 degree view, via just a shiny ball on top of a camera)
-- mofosyne, Jun 30 2010


The GPS receiver and 360 deg camera are pretty weak solutions because of the analog hole problem. They could be modified to spoof their signal by someone with enough technical expertise and determination. Communication with GPS sattelites is strictly one-way for civilian applications. A cryptographic authentication of your time and location would require some sort of two-way communication.

A much better idea would be if the DSLR camera broadcasted a hash to a distributed public network (eg. Usenet) as soon as a picture was taken, via mobile or satellite internet.

This would serve as proof that the picture, bit-for-bit, could not have been created after the broadcast. There still would be, however, no technical means to prove the location and exact time before the broadcast.

But this would suffice for some applications. For example, you take a picture of the World Trade Center collapsing at 9:59:00 am and the hash is published on 100 Usenet servers at 9:59:10 am. This would provide solid evidence that the image is not doctored. Why? a) Because the exact time of the collapse is public knowledge and it is impossible to photoshop a conspiracy into a WTC picture in just 10 seconds. b) Because you couldn't have predicted the WTC collapse and it's extremely unlikely that you would find a believable doctored picture that was created before the collapse, in those 10 seconds.
-- kinemojo, Jul 02 2010


Nikon have an authentication option on the D700 and some software that can be used to check authenticity. I have no idea how it works.
-- Ling, Jul 02 2010


Yeah, but Nikon's solution is dead easy to circumvent.

All you need to do is change the camera's settings and take a picture of your doctored picture.
-- kinemojo, Jul 02 2010


Here is another idea: the 360 deg camera is constantly recording a low res movie and the last 30 min are stored in the camera - a bit like a black box.

Every time you take a picture, the 30 min of video preceding it are stored along with the picture. Video is a lot harder to doctor than still shots.
-- kinemojo, Jul 02 2010



random, halfbakery