When can a photo be trusted?

Fourandsix on TwitterFollow @Fourand6
« We might not want you to buy FourMatch | Main | The JFK Zapruder Film »
Tuesday
Sep182012

Introducing Our First Product: FourMatch

I couldn’t be more pleased to announce that Fourandsix has now released our very first product: FourMatch. If you go to the FourMatch page on our site, you’ll find lots of information about the product, but I wanted to say a bit more about it here specifically for regular readers of our blog.

As we say prominently on the product page, FourMatch is a product that’s designed to “authenticate images instantly,” and that’s exactly what it does. It automatically identifies files that have been unmodified since they were first captured by a camera or mobile device. So, if you need to deliver strong evidence that a photo is truthful, FourMatch is the tool for you.

Last month, I wrote a blog post about using the “3 F’s” to determine the authenticity of a photo. FourMatch is a tool for examining the first of these F’s, the File. Unlike many of the techniques we’ve detailed in this blog over the past year, FourMatch does not examine the photo itself, so it doesn’t look for inconsistencies in the image. Tools to handle these other F’s—Footprints and Flaws—will come later. 

What’s special about FourMatch is something I hinted about in my “Null Hypothesis” blog post earlier this year. Most image forensics techniques are great for spotting signs of tampering in an image, but, when you fail to find one of these signs, you really haven’t quite proven that the image is real. It could just be an expertly created fake.

In contrast, if a photo passes the FourMatch test, you can be confident that it’s authentic. It’s when a photo fails this test that it’s less conclusive. That’s when some of the other analysis techniques—such as looking for shadow and reflection inconsistencies—can come into play. Thus, FourMatch can make a great triage step in a larger forensics workflow.

More importantly, though, FourMatch can solve a critical problem for some specific use cases. We expect it to be particularly valuable for users in law enforcement. Photographic evidence forms the core of some types of court cases, but doubts about image veracity have weakened the effectiveness of this evidence. With FourMatch, police and legal professionals have an important tool for establishing the reliability of this evidence.

In the coming weeks we’ll be blogging more about how to use FourMatch. But don’t worry, we won’t stop covering other types of techniques as well, and we have every intention of providing more tools for your analysis toolbox in the future.

PrintView Printer Friendly Version

Reader Comments (4)

Congrats guys!

Your first baby! You all must be very proud!

Would it be worth while marketing this as part of a tool set for digital archaeology? Allowing archaeologists the possibility of creating a "history" of an images usage and "derivation" (maybe even pinpointing the software used in those transformations??

regardless

Bravo!

[Thanks! We definitely plan for FourMatch to ultimately be part of a larger tool set. I'm not sure if the archaeology metaphor will be apt, because we won't necessarily be able to tell the complete history of the image from start to finish, but certainly we expect to be able to uncover important clues. - Kevin]

September 18, 2012 | Unregistered CommenterGam

Congratulations! Your first product sounds like something all press agencies should urgently install. Just curious, would a side-tour through Lightroom/Photoshop for processing the raw image count as processed, or straight from camera? (As long as no spot corrections are applied, well, it is not more processed than we did with negatives?)

[*Any* time the file has been re-saved, the signature will change, and it will fail the test. We intentionally take a very conservative approach with this. The example you've used is for raw files, but if someone wants to provide evidence that something is an original, simply providing a proprietary raw format is pretty strong evidence on its own, because it would be pretty difficult to fake that. An interesting tangent on your question, however, is the issue of using Photoshop's Camera Raw module or Lightroom to edit a JPEG file. Technically, you can bundle up all of your edits into the metadata of the JPEG without actually re-saving the image and re-compressing the file. Currently, this will still tend to trigger a failure in FourMatch, because the metadata will have changed, but you should still see that the dimensions, thumbnail, and compression are original. In theory, we could modify FourMatch so that this is special-cased, and these files could still pass. I could see this as being a viable workflow for instances where you want to allow some limited amount of editing. You're sending the original JPEG--which can be verified--but included in that file are all of the instructions for how the creator intends for the file to be edited. The receiver (e.g. a news organization) could then choose whether to accept or reject those edits. - Kevin]

September 19, 2012 | Unregistered CommenterAnn

Hi.
Maybe I misunderstood the purpose and the functions of this powerful tool, because some questions arise.

1.Postproduction is not an option in professional photography, but just doing some postproduction doesn't necessarily make a photograph a fake! Where is the limit to that?
Might simple postproduction be considered as manipulation by FourMatch?
2.What about RAW files? I mean, most photogrtaphers who shoot RAW have to export their files in jpg format (using Photoshop-like tools); everyone who want to submit a picture to an online competition has to resize it to the desired dimension (even if it was taken in jpg).
Would those operation result in "yellow" light (="could be a fake") or a "red" (="fake") one?

Thank You
Luciano

[Great questions. The purpose of FourMatch is to tell you if a JPEG file has come direct from a camera without having been re-saved or tampered with in some way. This is very useful when you need to produce strong evidence that the content of an image is "real." A proprietary raw file from a digital camera is inherently strong evidence on its own, because editing one of these files forces you to save it in a different format (just as editing a JPEG forces you to modify the signature).

It's also important to note that failure to pass the FourMatch test does not mean that an image is untruthful. It just means that the file may have been touched at some point, so you lose the certainty that you can trust it. If it fails the test, then you can follow up with further investigation, such as by using some of the techniques discussed in our blog. Ultimately, we plan to have additional tools that make this deeper analysis easier and more efficient, but for now a more manual approach will need to suffice.

There are certainly scenarios (such as photojournalism) where you'd like to permit modest amounts of editing (cropping, tonal adjustments, etc.) while forbidding more drastic edits. The best way to apply FourMatch to solving that problem is to require people to submit their original file along with the edited version. Then, you can compare both versions to make sure that the edits are acceptable, and you can also test the submitted original to verify that it is, in fact, an original. Admittedly, though, this workflow won't be appealing to everyone. - Kevin]

September 20, 2012 | Unregistered CommenterLuciano

so what is the answer to this...... http://www.whitehouse.gov/blog/2011/04/27/president-obamas-long-form-birth-certificate

i think i know...

[FourMatch is not designed to analyze scanned documents. It currently can only analyze images from digital cameras, mobile devices, and tablets. Having said that, we have previously reviewed the purported evidence claiming that President Obama's birth certificate is fake. We find these arguments lacking in substance and scientific fact. - Kevin]

September 22, 2012 | Unregistered Commenterrandall

PostPost a New Comment

Enter your information below to add a new comment.
Author Email (optional):
Author URL (optional):
Post:
 
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>