Thursday, November 21, 2013

NightThief Forensic Imaging Technology

I received the usual spam announcing a new company in the forensic imaging marketplace, FRP Forensic Imaging’s new NightThief forensic imaging technology.

The OCD in me was alerted by the specific use of several words in the piece - like technology vs. software or product and "restoration" vs clarification or enhancement. The OCD went into overdrive with this bit, "NightThief imaging is available as a controlled service built around uncompromising ethics and usage frameworks. FRP Imaging’s commitment to responsible forensic imaging includes strict submission requirements, chain of custody prerequisites, and a no-nonsense licensing model designed to facilitate smarter criminal investigations while protecting the privacy and liberties of innocent individuals." What's a "controlled service?" Submission requirements?

But then, the issue was laid to rest, "Because NightThief imaging is a service, there are no upfront costs, investments in time, equipment, training, or commitments required. More importantly, NightThief is built for speed with processing times rarely exceeding three days – and critical cases much faster! Dynamic all-inclusive per-event pricing based on senior agency involvement and community population provides simple, predictable, and affordable world-class forensic imaging to organizations and communities of all sizes." It's a service with per-event pricing. Ok. First question, who's going to come to Los Angeles to testify about the work done on the images?

So, I went to the web site - FRP Imaging. It seems that FRP stands for Forensic Reverse Pixelation. I couldn't find out what they mean by the term, but it seems that they're a tech start-up coming out of the movie special effects industry.

Judging from their FAQ page, they may run into significant resistance over a few of their policies.

  • "Does any other company perform FRP Imaging or use NightThief™ technology? No. FRP Forensic Imaging™ and NightThief™ technologies are exclusive services of FRP Imaging."
How will they handle the reliability/repeatability questions if they don't share their technology? What happens if they're compelled to release their tech to opposing counsel? What happens when their technicians are cross-examined about how their technology works?
  • "Who owns FRP-processed images? We do. To control the application or FRP images, we license copyrighted FRP-processed images to acquiring agencies. Agencies may use them freely within departments, and publicize them in association with their case specific law enforcement efforts in perpetuity."
I'm not sure how the "we license the results back to you - but we now own the file" policy is going to fly with LE agencies and the courts.
  • "Are FRP images admissible in courts? Generally, yes. However, it’s important to remember FRP Imaging products are designed as investigatory tools."
An investigatory tool that's generally admissible in court - at an "affordable price."

Back to the who's going to come to your court and testify about the work and be cross examined, as is the defendant's right? No one, it seems. 
  • "Each FRP processed image includes a notarized sworn statement made under penalty of perjury certifying the image was not modified, original enhanced, created, edited, or manipulated in any way that adds or removes any element from the image, or portrays any suspect in an unfavorable light beyond the actual image provided. If enhancement(s) were made, it is noted along with its purpose (scale, proximity, detail, etc.) ."
How do you cross examine a "notarized sworn statement?"

I'm interested to hear back from them to find out more about how they see this working.

1 comment:

Anonymous said...


I too wondered about the idea of pixel replacement. For example, you want to ID a license plate number but due to a lack of pixels, cannot get a positive ID. Now, they will "add" pixels to clarify the image. What is the methodology? Is there a chance of a false ID, for example, making a letter "E" vice an "F", or a "1" vice an "I"? Those are easy about making a letter or number not even close? I would like to see test images of known objects and see how often they get it right.