Featured Post

Welcome to the Forensic Multimedia Analysis blog (formerly the Forensic Photoshop blog). With the latest developments in the analysis of m...

Tuesday, April 28, 2015

New NVIDIA drivers will no longer support older CUDA GPUs

This just in from Adobe: "Just to alert you, NVIDIA is starting to discontinue CUDA support for some older GPUs in their new driver releases. If you are still using one of these older cards, and need GPU support for Adobe applications, do not update your drivers while you look into a purchasing a new card.

If you need to reinstall NVIDIA drivers, see this web page."

Wednesday, April 22, 2015

a little humor

A photon checks into a hotel and the bellhop asks him if he has any luggage.

The photon replies, “No, I’m traveling light.”

Monday, April 20, 2015

Police Cameras Bring Problems of Their Own

This just in from the Wall Street Journal: "As more police agencies equip officers with body cameras in response to public pressure, authorities are discovering they create problems of their own: how to analyze, process and store the mountains of video each camera generates.

Prosecutors in northern Colorado recently spent hours poring over a dozen videos captured by police wearing cameras. The case? An arrest for drunk and disorderly conduct.

Clifford Riedel, Larimer County’s district attorney, said his office has been overwhelmed with footage from the 60 body cameras the Fort Collins Police Department uses, and will need to hire an additional technician to sort through it all. “There are just huge amounts of data being generated from cameras,” said Mr. Riedel. “It used to be that video on a case was the exception. Now it’s the rule.”

The movement gained new intensity after the police shooting last week of a fleeing man in South Carolina. While many experts inside and outside of law enforcement agree that body cameras—clipped to officers’ uniforms or glasses—help increase police transparency and may even improve police behavior, police departments and prosecutors are struggling with how to sift through, preserve and share the visual evidence.

On top of that, agencies need policies and personnel to respond to requests from journalists and the public to release video under freedom-of-information requests.

“The vast majority of places are still trying to figure this out,” said Michael White, a professor of criminology at Arizona State University who wrote a Justice Department report on body cameras.

Dr. White estimates that between 4,000 and 6,000 U.S. police departments, out of about 18,000 nationally, use body cameras. Officers generally turn them on when stopping a driver or responding to an incident.

Some departments use body cameras in addition to dashboard ones that have become common at many agencies, but result in less-useful footage because much police action takes place away from their vehicles. Body cameras—which cost hundreds of dollars each—typically result in much more video for departments to handle.

The push to require body cameras intensified nationally after last August’s shooting of Michael Brown, a black 18-year-old, by a white police officer in Ferguson, Mo. This week, after a bystander’s cellphone video surfaced showing a white South Carolina policeman fatally shooting an unarmed black man in his back, several prominent state lawmakers voice support for a bill to require all officers to wear cameras.

But the cost has given some officials pause, said Lindsay Miller, senior research associate at the Police Executive Research Forum and co-author of a Justice Department report on the topic. “The cameras themselves aren’t overly expensive, but the years and years of data storage you’re going to deal with—that can definitely be cost-prohibitive,” said Ms. Miller.

Many departments keep inconsequential video for 30 to 60 days. But if the footage is evidence in a criminal case, it must be kept longer; most states require that video in a homicide case be kept indefinitely, she said. Ms. Miller said an emerging consensus is that the benefits outweigh the costs. In limited studies, the cameras have shown promise in reducing use of force by police and citizen complaints—and that can save money spent investigating complaints and settling lawsuits, she said ..."

Keep reading at WSJ.com.

Tuesday, April 14, 2015

Correct Fisheye Distortion in FIVE

I've seen a lot of ADT installations using fisheye lenses lately. ADT must be having a sales promotion or something.

One of the updates to FIVE in the last year is the Correct Fisheye filter, found in the Edit filter group.

In the past, I would use the Undistort filter, but Correct Fisheye works a lot better / easier / faster on these low-end fisheye lenses I've been seeing.


In the case of the "full-frame" (180º) fisheye lenses in the ADT installations, I've found that Orthographic works the best. This isn't because ADT has been forthcoming with information about the mapping function of their chosen lenses. It's mainly been through my testing each function. You won't likely get information on the manufacturer's chosen mapping function.


As a side note, for the "circular fisheye" top-down (360º) lenses, use the Unroll filter for better results.

Back to the ADT cameras, here's the before/after image (Presentation>Compare Original). On the left is the original image and on the right is the corrected version. Correct Fisheye did a very good job in just two clicks of the mouse.


Enjoy.

Monday, April 13, 2015

It looks like the transition to Amazon.com is finished and my book, Forensic Photoshop, is now available there exclusively. Paperback and hardcover options are both available for immediate shipping.

As I've noted in the past, the book isn't based on a particular version of Photoshop. It's a workflow book ... what to do when, and why. As such, it's still relevant and informative for those still using Photoshop in their forensic work.

Thanks again for your continued support.

Friday, April 10, 2015

What's next for Adobe Audition CC?

Adobe has begun to release details about the upcoming changes to their Creative Cloud products. Click here to find out what's coming next in Adobe Audition CC.

Thursday, April 9, 2015

A bite mark matching advocacy group just conducted a study that discredits bite mark evidence

Here's an interesting article from the Washington Post about Forensic Odontology.

Check out this quote, "the problem with bite mark analysis was never the lack of a flow chart. The problem is that there has never been any real scientific research to support its two main underlying premises — that human dentition is unique, and that human skin is capable of registering and recording that uniqueness in a useful way. And the research that has been done strongly suggests those two premises are not true." Ouch!

Read the whole article here.

Wednesday, April 8, 2015

Coursera courses reviewed


Last week, I finished the Coursera course, Visual Perception and the Brain. This course was taught by Dale Purves, MD, of Duke University and is one of the growing number of on-line non-credit courses offered by leading universities and noted professors around the world.

I would recommend that DME analysts put this course on their to-do lists and watch for it to be offered again. There's a lot of good information delivered as part of the class.

That being said, if you've never taken an on-line university level course you may have some trouble with the format. You watch the video of each section's lecture. You're free to download and save it locally. You can also download and save the slides and a fairly accurate transcript of the lecture. There's an expectation that you'll dive in a little deeper and study each section's topic on your own prior to taking each section's test. The test questions aren't written directly from the lecture / slides and assume that you've done a bit of extra reading in order to gain a deeper understanding of the week's topic.

Coursera has a lot of classes available for free. In this world of shrinking budgets and doing more with less, you can't beat free classes.

Monday, April 6, 2015

Recovering image data from JPEG file fragments

This interesting paper was recently published by SPIE. SPIE is the international society of optics and photonics.

"A new technique for recovering fragmented data files can retrieve elements of a JPEG compressed image even when the file's header is unavailable."

"The most basic task of any file system is to manage and organize data in a storage volume. Each file in the store is allocated a list of blocks (the basic unit of access), and when we access a file, the system retrieves the data in sequence from the list. Correspondingly, removing the relevant entry deletes the item (note that in most systems, deleting simply means that data is overwritten over time by newly saved files, rather than actually removed). Figure 1 depicts the layout of files distributed across blocks of a storage medium, and provides a simplified view of the data structure used to track allocated blocks. When the information for a volume is corrupt or missing, it is only possible to recover files by analyzing the fragmented raw data, a process known as file carving."

"Today, file-carving tools play an important role in digital forensic investigations, where analyzing deleted files and salvaging data from damaged and faulty media are common procedures. However, when files are encoded and compressed (as with most multimedia files), recovery is dependent on the availability of the file header, which includes all the decoding parameters. If a file is partially intact with its header deleted, common carving techniques cannot recover any data. Here, we describe an algorithm1 that advances the latest developments in JPEG carving by introducing the ability to recover file fragments when the associated header is missing.

There are two main challenges associated with file carving. The first is the inability to lay out data blocks contiguously on the storage, as in the case of files b and f in Figure 1. Repeated execution of file operations, such as addition, deletion, and modification of files, over time leads to fragmentation of available free storage space. As a result, the newly generated files need to be broken into several parts to fit into the available unallocated blocks. Figure 1 illustrates this phenomenon, where files a and c are separated into two pieces. Even the new solid-state drives (which use integrated circuits, rather than disks, to store data) are susceptible to this phenomenon as they are designed to emulate the interface characteristics of hard disk drives.

The second challenge to successful file carving is that interpreting binary-formatted data requires the use of decoders, without which a block of data will reveal little or no information about the content of a file.

There has been significant recent progress in carving JPEG files, which are the most widely adopted still image compression standard today and commonly the subject of forensic investigations.2–6However, techniques for JPEG recovery still assume that a file header is always present. Without the header, the usual techniques cannot recover any data, even though the rest of the file may be intact. For example, in Figure 1, the fragments of files d, e, and g were overwritten by files c and f. Recovering an arbitrary chunk of compressed image data without the matching encoding metadata essentially requires reconstructing a new file header, which at least requires knowledge of entropy coding parameters and quantization tables, methods used during compression for downsampling the color information and image dimensions."

Click here to continue reading this paper on SPIE's web site.

Friday, April 3, 2015

LEVA Job Classification and Wage Study Results

This just in from LEVA: "The following Job Classification and Wage Study was completed in March of 2015 as part of a position audit. The survey focuses on particular job duties and functions performed by various analysts, technicians, and those who in some way handle digital multimedia evidence.

All information contained in this report was volunteered by those who responded to the survey invitation, knowing that they fit the subject matter criteria. It is being shared with the Forensic Digital Multimedia Evidence community due to numerous requests for the information.

LEVA thanks those who responded with data for Jordan Huslig of the Grand Junction, CO PD Crime Lab to conduct the research and providing the findings. Nice job Jordan!"

Click here to read the result and/or access the raw data.

Thursday, April 2, 2015

How to: Open Exe Files on A Mac

I recently had an .Exe video file that wouldn't capture using any/all available PC tools. I tried Omnivore, FIVE, and a few others. Nothing. Thank God for my trusty old MacBook Pro.

Here's an old video tutorial that's still relevant. It demonstrates using Wine / WineBottler to re-bundle the file into something that the Mac will recognize.

Wine / WineBottler worked like a charm on my video. Then, I used iShowU to do the screen capture.

Sometimes folks can't/won't mess with executable files and require the files to be "converted." Not all of them play nice with our state of the art systems. It's nice to see that the old tricks still work.

Enjoy.