The Driving Force Behind Forensic Reconstructions

30 Forensic Engineering (30FE) is an employee-owned Canadian multidisciplinary engineering firm headquartered in Toronto with regional offices in Vancouver, Calgary, and Ottawa. The 97-person team works across 16 practice areas and is frequently hired by insurers and lawyers who want to understand their potential risk and exposure in the event of harmful incidents both in Canada and internationally.

Nishan Perera is the Practice Lead for 30FE’s Digital Analysis offering and a Senior Associate on the Collision Reconstruction team. He has handled over 600 cases involving pedestrian-vehicle impacts, multi-vehicle collisions, and, more recently, work related to video analysis, due to the rise of collisions caught on CCTV cameras and dashcams. 

“I’ve always been a problem solver, and figuring out how and why things work is part of my nature. Forensic investigations take that problem-solving nature and flips it to figure out why things failed,” says Perera, “SynthEyes is a game changer for those of us assessing collisions (amongst other types of cases). I’d go so far as to say that without this tool, our analysis would not be nearly as robust, and our understanding of the how-what-why’s of a collision would be severely restricted.” 

30FE: Side swipe case example

Solving Poor Quality Footage

“Unlike traditional VFX artists, most of the footage we get is low quality, comes from an unknown source, and depicts erratic camera motions,” continues Perera. “Before we even touch SynthEyes, understanding the limitations of footage is very important.” 

The first step is to investigate the metadata and inspect the footage to ensure that what is shown aligns with the reported frame rate of the video. The team then addresses any anomalies (such as duplicate frames) or areas that need enhancement before using SynthEyes. They also often use LiDAR data captured by 30FE’s FARO 3D scanner to ensure accuracy, given the sensitive nature of their investigations.

No two cases are the same, which allows Perera to explore the various options within SynthEyes to tailor the solution to the individual needs of each investigation. “Although the footage we deal with is complicated and less than ideal, the tracking process is very straightforward,” states Perera. “Getting to an accurate solution as quickly as possible is the most critical aspect of the process. I regularly utilize features that expedite tracking (such as key smoothing and stop on auto key) in conjunction with point cloud data integration. I also find using the phase editor helpful in certain cases where the camera motion is known to be linear (think of a camera on a streetcar).”

“SynthEyes allows us to not only speed up the process for tracking a moving camera, but because of its ability to track the camera position on a frame-by-frame basis expediently, we can generate high-resolution data of a vehicle’s motion, which accurately captures moments of acceleration or deceleration,” remarks Perera. “The onset of when these events occur can mean the difference between whether a collision can be deemed avoidable or not. Having reliable data allows us to be precise in our understanding of the collision circumstances.” 

After footage is tracked, the solved positions of the camera or objects are exported out of SynthEyes. Next, the team applies any necessary numerical filtering methods to smooth out noisy data and obtain the true speed profiles of the vehicles or objects that are the focus of the case.

30FE: Static Camera Vehicle example

Speeding Up Tracking with AI

Many car collision investigations involve traffic that may not be directly related to the case at hand. Due to this common occurrence, 30FE cannot routinely take advantage of the powerful auto-tracking feature in SynthEyes — until now. 

“By far, the Mask ML feature is one of the best upgrades in SynthEyes 2025. Auto-tracking is an excellent function in SynthEyes, but often unused due to the amount of other traffic we have to deal with in the footage, which would require roto masking before using the auto-tracking capabilities,” comments Perera. “The new Mask ML function allows us to very quickly ‘mask out’ vehicles we don’t want to track. It offers an excellent way to now integrate auto-tracking back into our workflow to generate an accurate tracking solution even faster.”

30FE: Use of force case example

A Valuable Teaching Tool

“When I first embarked on my journey to learn SynthEyes, it was overwhelming to say the least. Many of the tutorials available were geared towards problems a VFX artist may come across,” notes Perera. “It was difficult to understand the interface and workflow as they would apply to typical problems a forensic professional might face. That being said, it became very clear that SynthEyes was an essential tool to learn how to use in our line of work.”

Perera has developed a course to support the forensic community as a whole and make SynthEyes more accessible. Students will master photogrammetry theory and apply their knowledge using real-world examples that require the use of SynthEyes to determine subject heights or vehicle speeds through various tracking techniques. The course will be offered online (June 2-6) through the Law Enforcement and Emergency Services Video Association (LEVA), the global leader in forensic video analysis training. Registration for "Photogrammetry 1: PhotoModeler and SynthEyes for Forensic Video Analysis" is open.

“Teaching is the best part of my job. I take technically complicated scenarios and chaotic circumstances like a collision and distill them down to a simple yet scientifically accurate story about what actually transpired,” ends Perera. “Using a program like SynthEyes offers me the ability to provide that story in a visually compelling manner that just about anyone, regardless of their background, can understand.”