Laurence R. Penn, Senior Forensic Animation/Video Specialist
I bet you’ve seen those videos on social media with a friend’s face superimposed over an actor in a famous movie. If not, take a look at this video.
Recent advancements in video technologies such as artificial intelligence are making these funny, but worrying, deepfakes easily and quickly achievable by anyone with access to a smartphone or computer. It doesn’t take a huge stretch of the imagination to think of ways this could be used maliciously. Luckily, with expert knowledge in forensic video analysis, there are techniques to help root out the truth behind videos suspected of forgery.
Digital camera sensors have a unique hidden sensor noise pattern which isn’t noticed without filtering out various values of the image. Think of it like a thumbprint, it’s not noticeable until you look very closely, and it is unique to each individual. Over time, the sensor noise pattern should be consistent throughout the duration of the video. Any inconsistencies in the pattern, such as a computer-generated overlay of someone else’s face, would disrupt this sensor noise thumbprint inherently present in the original video clip, therefore making a convincing argument of tampering. Other techniques involve looking at changes in compression artifacts as well as inconsistent changes in motion or brightness. That being said, there appears to be no issues with motion or brightness inconsistencies in this pleasantly sunny video of “Tom Cruise” on the golf course…
This is not Tom Cruise
This is a deep fake
This probably isn't great news for Cameo pic.twitter.com/CqkzNOAFQp
— Damian Burns (@damianburns) February 26, 2021
Laurence R. Penn, Senior Forensic Animation/Video Specialist with DJS Associates, Inc., can be reached via email at experts@forensicDJS.com or via phone at 215-659-2010.
Richard A. Kennedy, QEI, CET, Elevator / Escalator Expert
Case Synopsis: A resident of a high-rise condo was descending on one of the elevators located in the building. When the elevator came to her designated landing, it decelerated normally and leveled into the floor where upon the doors opened to let her exit. While exiting, she claimed the elevator suddenly and without warning, fell downward causing her foot to momentarily get caught, which caused her to trip and fall forward hitting the floor and lobby wall.
Expert Analysis: Discovery documents revealed that the elevator in question had been modernized three months prior to the alleged incident, and contained all the requirements of the newest version of the ANSI 17.1 elevator code. Management of the condo association also maintained a log of all maintenance visits by the elevator company and documented any malfunctions reported to them by residents or guests. Management of the condo association checked the operation of the elevators when an incident was reported, and if there was a malfunction they secured the elevator out-of-service until the service company arrived and corrected the malfunction.
Most building owners do not employ anyone with the technical expertise to maintain or repair elevators and rely on the expertise of the elevator company. Plaintiff’s expert inferred that the condo association took a simplistic approach to solve the elevator problems and based his conclusion on the condo log had documented six visits, within three months, by the elevator company since the newly modernized elevator was in service. It was noted by the defense expert that four of the six logged visits were maintenance procedures. It was also alleged by plaintiff’s expert that the newly installed elevator had a history of bouncing and not leveling correctly. Documented evidence did not support this claim. When the elevator was inspected and turned over for public use, one of the control logic selections was for pre-opening of the doors as the elevator approached the designated floor in low leveling speed. This feature was disabled and documented in the adjuster’s manual. Plaintiff’s expert also claimed that the elevator stopped level with the floor and then dropped four to six inches. The new version of the elevator code stated that the elevator had to be fitted with a rope gripper to prevent unintended movement, which it was.
Case Result: Defense Verdict
Richard A. Kennedy, QEI, CET, Elevator / Escalator Expert with DJS Associates, Inc., can be reached via email at experts@forensicDJS.com or via phone at 215-659-2010.
Tom Griffiths, Ed.D., Aquatic Safety Expert
In a large 50-meter swimming pool, an expert swimmer in the military drowned while “dolphining” for extended periods of time. He worked out strenuously on the surface prior to his underwater exercises, so it is likely that he depleted his oxygen and carbon dioxide stores prior to his underwater swims. This was a classic Shallow Water Blackout scenario, where a good swimmer overexerts themself in the water and then holds their breath for long periods of time repetitively underwater.
Dolphining is an underwater kicking exercise where the swimmer, wearing fins, will swim many underwater lengths while performing an up and down dolphin kick. During dolphining, the expert swimmer fell unconscious and the lifeguards on duty missed recognizing him underwater, which is not difficult to do. When someone quietly slips beneath the surface, the water hides and suffocates its victims. In addition, lifeguards tend not to watch these underwater swimmers because they are such superior swimmers, often stronger swimmers than the lifeguards on duty. Lifeguards likewise wrongly believe good swimmers do not drown.
In this case the family of the deceased blamed the lifeguards for their inattentiveness. Conversely, the defendants blamed the underwater swimmer because he should have known competitive and repetitive underwater swimming is dangerous and deadly.
Lessons Learned: ALL swimming pools MUST ban extreme underwater swimming and prolonged breath-holding. Lifeguards must be trained to enforce this vitally important rule.
Tom Griffiths, Ed.D., Aquatic Safety Expert with DJS Associates, Inc., can be reached via email at experts@forensicDJS.com or via phone at 215-659-2010.
James R. Schmidt, Jr., BSME, Collision Reconstruction Engineering Analyst
Okay, if it’s not speed from video that I’m writing about, then it’s sun glare!
Timothy R. Primrose, Mobile Forensic Analyst
Apple’s health application monitors and documents a user’s day to day activity in detail. Some data is collected by the iPhone, while other data can be collected from a synced Apple watch or input directly by the user. These datatypes include steps, body measurements, menstrual cycle tracking, hearing, heart rate, mindfulness, mobility, nutrition, respiratory, sleep, vitals, and other user input data such as blood glucose and inhaler usage.
From a forensic standpoint, you might not think that health data can help decipher the events that unfolded in a given situation; however, some of these datatypes can be utilized. The hearing datatype proved useful in one of our recent cases.
The purpose of the hearing category in Apple’s health application is to monitor headphone audio levels to ensure that the user’s ears are exposed to a healthy level of sound decibels to prevent hearing issues caused by extended exposure to loud audio. In the image below you can see the audio levels that a user’s headphones were set to, the timeframe that they used their headphones, the media source of the audio, and even the type of headphones utilized. Users can view their listening history that Apple has recorded within the health application.
How was hearing data useful in a forensic investigation? A vehicle operator was traveling at the posted speed limit of 45 mph when a woman suddenly stepped out in front of him. He did not have enough time to stop and unfortunately the woman did not survive. The woman had headphones in and was reportedly looking at her phone when she stepped into oncoming traffic. Data extracted from her cell phone showed that the pedestrian had the YouTube application open leading up to and at the time of the incident. Audio level data recorded by Apple’s health application showed that the user’s device was outputting an audio level of 150 dB. This volume level would have prevented the woman from hearing any approaching vehicles or the buzz of traffic. Seeing that the YouTube application was open and not an application such as Pandora or Spotify aids the vehicle operator’s story that the pedestrian was looking at her phone and distracted from where she was walking.
Health data can be utilized for more than what Apple intended. Even if you do not plan on using this information for a case, check it out for yourself and make sure your audio levels are safe.
Timothy R. Primrose, Mobile Forensic Analyst with DJS Associates, Inc., can be reached via email at experts@forensicDJS.com or via phone at 215-659-2010.
Robert T. Lynch, PE, Principal Collision Reconstruction Engineer
A colleague recently sent me a link to an online game to test one’s reaction time (see link below). The game simulates a (right-hand drive) vehicle traveling down a straight, hilly roadway with trees, signs, and deer to the sides of the roadway. When the player sees a red stop sign with a hand on it, they are to click the mouse or press any key on the keyboard to stop the simulation. Reaction time, the time between when the stop sign is presented on the screen and when the mouse or keyboard is clicked/pressed, is then compared to how 2,000 people above the age of 18 performed in this same simulation.