I learned recently that computer animation, and modelling or simulation, while useful in forensic work, including accident reconstruction, can also be misused. Sometimes by mistake, other times deliberately.
It doesn’t help that software sellers claim accuracy but don’t give data backing up their claim. Nor do the claims that animation can help “build a solid case” when forensic experts must be objective in serving the judicial and dispute resolution processes.
I wonder to what extent each of you are aware of these problems:
- Insurance claims adjusters
- Civil litigation lawyers
- Judges and jurors
- Dispute resolvers
- Injured parties, in general
If you’ve got a few moments, the following is certain to enlighten you.
To animate means, figuratively, to “give life to”. For example, make an object seem to move. The moving cartoon figure seen in a TV commercial or a teenager’s cell phone might begin as two inanimate pictures of a figure with the head or an arm in different positions in each picture. Animation software adds additional pictures between the two with the head or arm in slightly different positions. This gives the appearance of movement to the cartoon figures.
A distinction must be made between animation which simply moves objects between specified start and end positions and modelling or simulation which does this based on characteristics of the object.
For example, in modelling the characteristics of a vehicle like a car might be it’s weight, tire friction, and how it brakes and steers. In animation these characteristics might not matter so much.
The object in forensic work might be a car involved in an accident that an expert is reconstructing to determine cause.
The expert collects data from the accident scene like the road conditions at the time, the location and length of skid marks, car specifications and impact speeds.
Animation and modelling software allows the reconstruction expert to place this data on a Google earth image or on a photograph taken from a drone or an airplane. The expert can also add data like labels, text and direction of travel.
If the topography of the site and the location of things on or beyond the road have changed since the aerial photographs were taken, the software allows the expert to correct this data.
In the old days this would be done by a draftsman drawing the scene from a land surveyor’s notes and adding the data from the reconstruction expert. Slower for sure but also more accurate.
After getting this input the animation/modelling software provides a set of data which describes the motion of objects at the accident scene – not unlike the cartoon figure in the TV commercial. For example, the speed and direction of travel of the cars involved in the accident.
The reconstruction expert can correct the software input data based on evidence from the scene if the output from the software is at odds with the field data. For example, the location of skid marks or the location of damage on the car.
This would be good use of animation software – as long as the accuracy of the software is understood compared to actual measurements at the scene by an expert and/or land surveyor.
Examples of good use
Following are some examples of the good use of animation and modelling software, and supporting techniques like Google Earth and aerial photography:
- As indicated above, correcting the input data to the software based on evidence from the scene – taking into account the accuracy of the software
- Input honest data to the animation and modelling software, rather than tweaked data designed to produce a desired but misleading result
- Checking the accuracy of animation and modelling software used in accident reconstruction At the very least, query the seller about the basis for their accuracy claims. Better still, check using independent data from an accident site or a failure in the built environment. This should be done before or during use of the software Do this for sure during peer review of another expert’s report on forensic work that relied on computer software Example: I read a report one time on an accident reconstruction. The speed of the vehicle was an issue. The software gave a speed that was greater than a verbal report by one of the parties. An independent check using different data confirmed the software output and the incorrect verbal report
- But experts and their clients must be careful. Accident reconstruction using a Google Earth picture of a site in an urban area can be very reliable because the resolution of urban Google Earth pictures is good. That of sites in rural areas is poor. This is because Google Earth pictures in urban areas are taken at lower altitudes. Example: I analysed the cause of a retaining wall failure on a residential street in Ottawa using an image from Google Earth. The resolution was so good you could take off the size of cracks in the wall. This kind of accuracy would approach that needed in accident reconstruction using animation and modelling software.
- The source of the site image used by animation software is of interest too. Site maps and images based on laser scanners (check Lidar on Wikipedia) are excellent. Unfortunately, they are less available. Example: I investigated the cause of a swimming pool failure in Cape Breton a few years ago – excessive foundation settlement. But why? The site was in a rural area that I was surprised to find had been scanned by a laser. I learned that it was a trial use of this mapping technique – lucky for me and my client. I studied the laser picture – a task called terrain analysis in civil engineering – and saw that part of the swimming pool was built over wetland, a swamp. This was not evident on the ground. Example: On another occasion I investigated the cause of fuel oil contamination of a rural site up on the Cape. In preparation for terrain analysis, I saw that Google Earth imagery was too blurred, the contours on topographic maps were too large, and laser scanned imagery was planned for this very area but not done yet. So I had video taken from a drone of the contaminated site and that solved an important issue at the contaminated site.
You must be careful about the aerial pictures used in animation and modelling software same as you must be careful about the claimed accuracy of the software. Honest input to the software, understanding image resolution, and investigating software accuracy are good uses of animation software.
Surprise! A hired-gun-expert could input data from the accident scene but tweaked a little to support building a solid case. This would be naughty use of animation software.
Examples of naughty use
Following are some examples of misuse of animation and modelling software by mistake or design:
- Accuracy of a vehicle’s motion. We know that any velocity, distance, orientation and time can be input to software and then set to motion and recorded as video. Example: In animation, incorrect, even impossible values can work. Examination by an opposing expert should be able to find these errors.
- Use of animation offered by current software that has realistic, unverifiable effects such as lighting, surface textures, vehicle damage, human figures and colours. The problem arises when the effects are inaccurate, producing video and stills/screen grabs that can prejudice a viewer, without the viewer even being aware of it. Examples of tricks you can play with software that’s on the market now: (a) Colour and lighting can increase or decrease contrast, causing a human figure representing a pedestrian to appear more, or less visible, or conspicuous. (b) The sky effect can make a nighttime scene look darker or lighter. (c) Vehicle headlamps, tail lamps and brake lights can be more or less conspicuous than they actually were. (d) The light pattern from animated headlamps may well be more or less than produced by the actual headlamps. (e) The colour and texture of road surfaces can leave a viewer with the subconscious idea that the surface was slippery. (f) Vegetation (bushes, trees, etc.) can be more or less of a view obstruction/restriction than the actual vegetation.
- Animations need to be carefully authenticated for accuracy (both motion and visually) before they are admitted into evidence. The probative value of evidence needs to outweigh the prejudicial. Example: The problem can arise when the court is not aware of how, or how much an animation is prejudicial. That’s when the expert needs to inform their client.
Animation and modelling/simulation software are wonderful tools for those of us doing forensic work. But using effects that cannot be authenticated must be kept to a minimum. Some skillful animators can’t seem to resist, whether it’s to produce a pleasing result, or a misleading one, maybe both, intentionally or unintentionally.
(In the spirit of Wikipedia, these two sections of my blog can always use additional examples of good use and naughty use of animation and modelling software)
I’m sure this issue of animation in forensic work is new to a good many non-technical clients representing injured parties. What’s an injured party or a claimant to do surrounded by such technology?
Here’s what you do. Ask questions of the forensic experts about
- the use and misuse of the animation and modelling/simulation software,
- the accuracy of the software,
- the shortcomings of the software, and
- be on guard and ever alert to the deliberate misuse.
- And get answers in jargon-free language.
It’s not hard to understand that you can play games with animation in forensic work not unlike what’s done in TV commercials and on kid’s cell phones.
The content of this blog is based on my forensic engineering experience, research online, discussion with others in forensic work, and common sense.
(Posted by Eric E. Jorden, M.Sc., P.Eng. Consulting Professional Engineer, Forensic Engineer, Geotechnology Ltd., Halifax, Nova Scotia, Canada, Updated: January 20, 2022. firstname.lastname@example.org)