Joshua Cohen is a principal at Fat Pencil Studio
When I saw the headline in Monday’s Oregonian, “Judge throws out 3d model,” I was nervous.
The case in question, USA vs Joseph Astarita, has generated tremendous media interest because it is related to the shooting death of Robert “LeVoy” Finicum, a spokesman for the group of armed protestors that took over Oregon’s Malheur National Wildlife Refuge in 2016. Astarita is not charged with firing the shots that killed Finicum, but with lying to investigators about discharging his weapon. This week’s news is not even about the actual trial, but rather a pre-trial hearing to determine whether certain expert testimony can be shown to a jury.
And yet, I’ve been following the case carefully because the methods used by Toby Terpstra (the government’s 3d animation and reconstruction expert) are very similar to those we use at Fat Pencil Studio. Will US District Judge Robert Jones’ ruling in such a high profile case make it difficult to have 3d models admitted in future trials? Let’s just say I was one of the first people to download and read the full details.
Like Terpstra, I’ve been in the hot seat to explain the basis and methods for creating 3d visualizations of crime and accident scenes. Earlier this year, I used a 3d model while testifying in a Portland police shooting case. The evidence included an aerial surveillance video that showed the location of officers, and brief flashes of light presumed to be muzzle flashes associated with gunfire. I encountered the same issues as Terpstra regarding an unknown focal length of the surveillance camera, which allows us to find line-of-sight, but not an exact location of the airplane camera. However, in this case, I only needed to explain that the flashes could not possibly have come from the plaintiff’s gun. Photogrammetry and 3d visualization techniques helped me explain to the jury why this was true. At the time shots were fired, the plaintiff was not visible to the aerial camera– it wasn’t even close. Any uncertainty around the precise location of the muzzle flashes was far outweighed by the long distance to plaintiff’s location. In other words, accuracy is relative. Getting an exact answer is less important than whether the accuracy is sufficient to answer the question posed.
In a Las Vegas wrongful death lawsuit involving a bus and a bicycle, we used security camera footage to help pinpoint the location where the collision occurred. As in the Astarita case, the item of interest (in this case a bicycle) appeared only as a “fuzzy smudge” in the video stills. However, we were aided by an additional bit of physical evidence—a black streak on the side of the bus that all sides agreed was caused by impact with the bike handlebars. I used a 3d model in court to show how knowledge of the bus path (which was accurately reconstructed using photogrammetry) could be combined with physical evidence to infer the location of the collision. I also used the model to show jurors possible locations of the bicycle a few seconds before the crash, and how this would look from witness perspectives. However, I made it clear at the outset, when describing 3d visualization methods, that considering the foundation of what you see on screen is of critical importance. Some elements, such as the roadway and bus path are based on scientific methods such as 3d scanning and photogrammetry. Other elements such as the position of the bicycle prior to the crash may be based on eyewitness testimony or a hypothetical scenario offered by an attorney. All of this can be helpful in understanding what happened, but careful thought must be given to the weight of the evidence. Jurors are smart enough to make this evaluation for a 3d model, they just need an honest introduction to what they are seeing.
In the Astarita case, Judge Jones heard many hours of argument regarding photogrammetry (or “camera matching”) methods and errors that may have affected the accuracy of the reconstruction. In Monday’s ruling, he largely sidestepped the issue of whether Terpstra’s methods were reliable, and instead focused on the human figures shown in the model, finding that the location and body position could not be reasonably inferred from the fuzzy smudges in video evidence: “To present it to a jury would suggest a degree of certainty that cannot be justified and would be unduly prejudicial under [Federal Evidence] Rule 403.”
Judge Jones did leave the door open for the government to use a 3d model of the scene and vehicles as context for eyewitnesses to describe the position of human figures. How will the government adjust their strategy going forward? How will others choose to interpret Judge Jones’ ruling in future cases? I’m not good at predicting the future, but I will suggest (to any attorney that will listen) that it’s very important to consider whether an expert’s method of inquiry has sufficient accuracy to provide a definitive answer to the question being considered. This issue could benefit from another article, but for now I’ll conclude by suggesting that if the answer is not crystal clear, it’s in your best interest to refine the question or adjust the presentation strategy.