Forensics & Science Case Studies
Scanning a WWII German Mini-Submarine with Artec Eva
Summary: A sunken German mini-submarine was discovered in a river in the Netherlands. Specialists decided to scan it in color 3D for historical preservation.
The Goal: To scan the hull, propeller, and both torpedoes of the submarine in high-detail 3D using a lightweight structured-light scanner. Then to create a final 3D model of the sub.
Tools Used: Artec Eva, Artec Studio
3D scanning technology is used to restore and preserve the smallest submarine of the navy of Nazi Germany.
Who can tell how much heritage has been destroyed by time, man and nature? Countless monuments and artifacts have been lost forever, leaving traces only in manuscripts, books, photos and memories of the lucky ones to have seen them. But things have changed with the advent of 3D scanning as more and more institutes and museums have started to embrace the technology to save precious and fragile legacy in 3D.
Artec 3D scanners have been used extensively to digitize museum collections and historical sites, from scanning Assyrian reliefs at the British Museum to 3D capturing excavation sites with fossilized bones of prehistoric animals and hominids in Kenya.
The extent of heritage preservation is not confined to antiques and fossils, though. Artifacts of more recent history also need to be conserved. One of the examples of such artefacts is this Biber mini-submarine that has been 3D scanned by Artec’s Dutch partner Erwin Kanters, the head of the 3D tech company Miniyours.
Scanning the hull with Artec Eva. Photo: Roland Blok, Stichting Maritiem Erfgoed K-Verband
The Biber, the German for beaver, was the smallest, one-man submarine in the Kriegsmarine, the navy of Nazi Germany. It could carry mines and two torpedoes attached to each side of the hull and was used to attack ships off the Belgian and Dutch coasts during World War II.
The submarine was developed in less than six weeks in February 1944 with the view to helping repel the imminent invasion of the Allies. The haste resulted in some serious construction flaws, which meant that pilots operating the Biber were virtually on a death mission. Between January and April 1945, 109 Bibers were sent on operations and only 32 survived.
The Biber scanned with Artec Eva sank before it saw action. The submarine was discovered buried deep in the silt of a river in the Netherlands a few years ago, and the Dutch maritime heritage foundation Stichting Maritiem Erfgoed K-Verband asked Erwin Kanters to 3D scan it in order to capture its measurements for restoration and preservation.
“Because all the metal was old and dented, it was quite easy to scan as we had enough features on the surface of the metal,” Erwin says.
Scanning the hull with Artec Eva. Photo: Roland Blok, Stichting Maritiem Erfgoed K-Verband
Using Eva connected to a battery pack, which ensures stable scanning in places where there is no source of electricity, Erwin scanned the hull, the propeller and the torpedoes. Even though the Biber is a mini-submarine, it was quite a large object for the scanner, and taking into account all the ribs and gaps that needed to be captured, scanning took one full day, generating 10 files of raw data 3GB each.
Scanning the propeller with Artec Eva. Photo: Roland Blok, Stichting Maritiem Erfgoed K-Verband
“I found it useful to make multiple files and scan the submarine in sections to get maximum accuracy, as each section took quite a number of scans,” Erwin says. “I find Artec Studio very fast and easy to use. Although I prefer to do post-processing manually, the automated post-processing is simply amazing! Artec Studio algorithms help a lot in processing. I especially like constrained alignment with loop closure.”
Part of the hull in Artec Studio
The propeller in Artec Studio
Erwin processed the propeller and parts of the hull, and the larger part of processing was done by the foundation’s staff. They are now reconstructing the Biber, planning to put it on display in the future, which may take some time as it was badly damaged.
3D Scanning For Traffic Accident Reconstruction:
Challenge: To 3D scan crashed vehicles for accident reconstruction & investigation, capturing the full extent of body damage in color submillimeter 3D, for in-depth inspections, analyses, and documentation, with the resulting data and conclusions being admissible for legal proceedings and court.
Solution: Artec Leo, Artec Studio
Result: Vehicle scans that used to require 3-4 hours, lengthy hardware repositioning, and extensive software post-processing with a terrestrial laser scanner now take less than one hour, using a professional wireless handheld 3D scanner that delivers high-resolution geometry and photorealistic textures of vehicles including all crash damage.
Scanning a wrecked 2014 Dodge Charger with Artec Leo
Every time forensic accident reconstruction expert Jarrod Carter, Ph.D., steps up to a twisted wreck, he sees a book of stories waiting to be told: how fast was the car moving when it slammed into the bridge? How many seconds before impact did the driver hit the brakes? Was everyone inside wearing their seat belts? And how well did the car’s safety features protect its occupants?
When Carter and his team are called upon to help tell the story behind a wreck, they collect a wide variety of data, including 3D site models from drone photos and tripod-based laser scanners, police reports, scene photos or videos from police or others, surveillance footage, stop-light camera video, dash cam video, and event data recorder (EDR, a.k.a. the car’s “black box”) data.
These varied sources offer up a broad spectrum of details from the seconds leading up to and including the crash: brake usage, accelerator application, steering wheel angle, lateral/longitudinal acceleration, roll rate, engine RPMs, gear positions, and more.
A recent addition to their storytelling toolbox is the Artec Leo.
Origin uses the Leo, a wireless handheld 3D scanner unlike any other, to tell the story behind the twisted metal that is a wrecked vehicle. Based on experience, they have found that the Leo can faithfully generate a digital twin of a vehicle’s exterior or interior in under an hour, from bumper to bumper, with submillimeter accuracy.
Artec Studio software screenshot showing the Leo scan of the 2014 Dodge Charger
In the past, as part of their push to create digital twins of wrecked vehicles, they used a tripod‑mounted 3D laser scanner. The scanning process entailed physically repositioning the scanner numerous times around the vehicle, at multiple elevations, inside and out, to record as much detail as possible.
And, even with all the repositioning, the detail, while tremendously better than the plumb bob and tape measure method Carter used at the beginning of his career, was still lacking in the qualities necessary to create a convincing digital twin.
Another significant problem with the tripod-based laser scanner workflow was time. Each scan with a tripod-based scanner takes several minutes, not including the time associated with repositioning. A detailed scan of a vehicle could easily take an hour and, in some cases, as much as two or three hours.
A brief window of time to scan the entire vehicle
When Carter and his team get down to work, they generally have a window of four to eight hours of access to a vehicle, and scanning is not their only agenda item. And, as a rule, they treat their time with the vehicle as though it is the last time they, or anyone, will ever see it. So, any time savings afforded during the scanning phase provides extra breathing room to ensure that the inspection is as complete as possible.
“This (the issue of time) is one of the main reasons I kept my eye out for a better solution than our tripod‑based laser scanner. I was looking for speed and flexibility, which Leo gives us, especially since it has no cables or attached computer to slow you down. I no longer feel as though the rest of the inspection is being rushed so that I can make sufficient time to scan the vehicle,” Carter said.
Scanning a Dodge Minivan with Artec Leo
“Because the vehicle is at the very center of our work, it is important to spend enough time gathering the data needed for its digital twin, even when that took us a lot longer than it does today. Now, with Leo, we collect the data for the vehicle’s digital twin so much faster. And I can use the touchscreen on the back to check the quality of the 3D mesh it’s making, the texture being captured, to make sure I have what I need before I leave,” said Carter.
“If I missed some aspect of the vehicle or didn’t get the detail I wanted in an area, I can easily rescan the bit I’m interested in with a wave of the scanner. It’s not like in the past with our tripod-based laser scanning workflow, where we’d have to wait till we got back to the office and started processing the data before we realized that some texture or geometry was captured less than ideally. With Leo, when we walk away from the vehicle, we’re confident it’s all there.”
The crucial need for extensive, true-to-life textures
As it relates to the texture data, Carter expressed high praise for the Leo’s capability, “What we weren’t expecting with Leo was the fidelity of the texture information it captures. The color and surface details appear photorealistic, or very nearly so. And the texture is not isolated to the individual points in the point cloud, like with a tripod-based scanner.”
“Instead, the texture fills the spaces between those points. A side benefit of filling in the gaps comes when you examine the vehicle from perspectives you didn’t consider when you were at the inspection. Now I’m not limited by the photos I took at the inspection. I can generate, on demand, what look like inspection photos of different aspects of the vehicle.”
Carter continued, “With Leo, we get highly accurate scans that provide more than enough geometry data for any analysis we need to conduct. And then you add in the photo texture to make it real. I remember the first time I zoomed in on the model of a vehicle we captured with our Leo. It was phenomenal. It looked exactly like the vehicle, which is what you want with a digital twin. We weren’t able to generate such high-fidelity models in the past. Not even close.”
Artec Studio screenshot showing the Leo scan of the Dodge Minivan
Having a digital twin of the vehicle that’s lifelike down to the smallest detail has become an expectation for Carter and his team since they started using their Leo. “When we sit down to go through the Leo scans in Artec Studio, it’s like being right there alongside the vehicle exactly as it looked during the inspection. We can visualize the evidence from any perspective we want, and we can measure it with exceptional accuracy,” Carter said.
Inspecting vehicle damage in Artec Studio
Carter explained one way they use the data from their Leo, “Once we compile the scans of a damaged vehicle, we align any undamaged portions of that vehicle with a 3D model or scan dataset of a similar undamaged vehicle. The comparison of damaged to undamaged allows us to determine the extent of crush on the damaged vehicle, which then provides a springboard for determining the direction and magnitude of collision forces, as well as how much energy was absorbed in the collision.”
“We can estimate change-in-velocity (delta-V) from absorbed energy and impact speed with enough other evidence. Additionally, we can use the comparison between the damaged and undamaged vehicles alongside our collision analysis to assist other experts who are trying to determine how the occupants were injured, and still other experts who are assessing the potential that some aspect of the vehicle’s design or manufacturing caused or enhanced those injuries.”
From 3D scan to biomechanical injury analysis
Origin Forensics also uses the data from Leo for biomechanical injury analysis. Here Carter and his team translate the outer crash event to the events involving the occupants inside the vehicle. A key aspect of the analysis focuses on determining how the occupants were interacting with the interior features of the car from the moment of impact onwards.
In Carter’s words, “We match up the occupant’s injuries to the elements of the passenger compartment that caused them, and determine whether any of the safety features there failed to perform as expected, whether that’s airbags, seat belts, or something else that was designed to mitigate injury. Could something have been designed or manufactured differently to prevent those injuries? We analyze every possible scenario, from beginning to end.”
Carter and Rothwell reviewing Leo scan data of the 2014 Dodge Charger in Artec Studio
During initial consultations with a client, in-person or over the web, Carter can share his screen and bring up the digital twin his team generated with their Leo, pointing out and explaining any relevant details.
Inspecting the Charger’s crush deformation patterns using Artec Studio’s surface distance mapping feature
As Carter explained, “Navigating around such a detailed 3D model provides a valuable adjunct to any 2D photos of the vehicle, which are frozen in time from the chosen perspective. Often the client will be curious about a specific aspect of the vehicle and we can take them right there and show it to them as though we were standing next to the vehicle or looking at a photo taken from that particular perspective.”
Using Leo scans for comprehensive vehicle damage reports
Following initial consultations with a client, there may be a need to submit a written report or to testify in deposition or trial. Generating exhibits that help the report reader or jury member understand the nature and extent of the damage sustained by an involved vehicle, or vehicles, is frequently an integral part of the process. And the 3D models generated from Leo scans provide the assets needed to create compelling visual exhibits.
Forensic Technician Kyle Rothwell, Origin Forensics’ in-house expert on Leo, described how he processes the Leo scans in Artec Studio software: “After importing our Leo scans, first I run Global Registration on a group of scans, then Outlier Removal on each of the groups, after which I align them.”
Rothwell processing the Leo scans of the 2014 Dodge Charger
“Then I clean up any stray geometry data, such as bits of glass, dirt, asphalt, etc. Once the raw data is registered, aligned, and cleaned up, I orient the scans to set the ground plane and rotate the object so that the right side view = the right side of the vehicle.”
Rothwell continued, “Then I run a Sharp Fusion, followed by a Fast Mesh Simplification. For a vehicle, a mesh density of about 2 million to 5 million triangles is appropriate for what we need. From there, I will apply the texture information for export and select the reduced glare. I normally use an 8K texture map to retain the smaller details. Then the model is ready for export, usually in .OBJ format with .PNG texture.”
3D digital twins of vehicles: even better than the real thing
Even though the majority of the cases that Carter and his team handle are settled out of court, or dismissed, and therefore never go to trial, if they do, their Leo has given them the ability to do what they always dreamed of: to put a true-to-life, virtual representation of the vehicle right in front of the jury.
“It’s even better than being up close with the vehicle itself, since with the digital twin, I can zoom in, rotate it however I want, and show everyone any part of the damage from any angle or magnification they’d like. And all of this evidence is fresh from the accident, so it represents what I saw during my inspection,” said Carter.
Preparing for inspection in Artec Studio: merging the Charger scan with an exemplar 3D model of the same vehicle
“In the not-too-distant future,” Carter hinted, “it may be commonplace that juries will have their own monitor or be wearing VR goggles when such exhibits are presented, which would make the impact of the Leo scan data all the more unforgettable. We could take them on a guided tour around or inside a vehicle, calling their attention to key aspects on demand.”
Using Artec Studio’s surface distance mapping to visually inspect the crashed Charger’s damage
“In the past, we would have to bring the actual car to the courthouse and take the jury out to see it, which is an expensive operation with no guarantee the court will even allow it. With the data we generate from our Leo, now we can bring the car to the courtroom and let the jurors walk around it virtually.”