Not Just Pretty Pictures…

Digitization is no longer just a novelty – it is absolutely vital to good analysis and communication of data, and it’s so easy more people should be using it.

Most people reading this blog will probably know what photogrammetry is, and it’s not my aim here to discuss the method per se. Suffice to say that with a digital camera (say, the one on the back of your smartphone), you can take a few pictures and then throw them into [free!] software that runs on a laptop, and produce a highly accurate 3D digital model.

Photogrammetry had been around for some time. Among others, my colleagues Brent Breithaupt and Neffra Matthews had been pioneering its use on dinosaur tracks for some years. But until around 2010, the good photogrammetry software had been about the same cost as a low-end laser scanner, and required a fair amount of user intervention, making it pretty much as inaccessible as laser scanning to most people, which meant 3D digitization was relatively rare in the literature.

That changed around the turn of the decade with fast, free, automated photogrammetry software. All that was required was a folder with photos in, and an executable that was double clicked. I reported on this in 2012, at which point the technology was moving so fast that between me submitting my paper in late 2010, and that paper being published in January 2012, the algorithms had moved to the GPU (i.e. the graphics card) and we saw a speed up of 100x. [By ‘we’ I don’t include myself, but rather refer to the extremely clever computer vision experts that actually write the software].

I really thought this would usher in a wonderful new age of data sharing, analysis, and communication.

To some extent it did, but not to the extent I had hoped. I think that’s because some people still don’t see digitization as anything more than making pretty pictures, or think that it’s hard to do. It’s neither of those things, and I’ll explain why.

Communicating and presenting data

Ichnology, perhaps even more so than many other areas in palaeontology, suffers during the transition from complex three-dimensional morphology in reality, to the limited two dimensions of print (be that paper or normal pdf). A track is a highly complex shape, and every aspect of that morphology is tied to some aspect of its formation.

Traditionally, tracks have been presented in papers and books as outlines. This practice goes back to the very beginnings of ichnology some 150 years ago. Outlines record the ‘vital statistics’ as it were – length, width, interdigital angle, digit thickness. But there’s so much more to a track! The areas that get raised up and pushed down can tell us far more about how the foot was moving than any plan view:

And because a track is the result of foot anatomy, substrate, and foot motion, that’s important! Someone that wants to look at the way the foot moved, or how the substrate reacted is going to be unable to use data that’s only presented in a 2D top-down image.

Recording specimens before they are destroyed.

Sometimes, particularly with tracksites, it’s impossible to collect a specimen. That means that a specimen that might be quite important (a holotype, or an unusual morphology) has to be left to the ravages of weathering, erosion, and vandalism. When future researchers come to build on the work that reported it, or to verify the original description, what is observable may be altered from how it was originally seen. Recording everything in 3D captures all of the morphology, and with a good enough model, or even good photos for someone else to build a model, all the information that was available to the people that discovered/described the specimen can be made available to other workers even if the original describers didn’t see or notice it.

Recording specimens after they’ve been destroyed.

I won’t say too much here, as fairly recently my colleagues and I had a paper out on the matter. Basically, if you have multiple photographs or video footage of a destroyed specimen, chances are it may be possible to recreate it digitally.

Do it!

Clearly, there are serious benefits to digitising specimens as part of a documentation and publishing procedure. So why isn’t it standard yet? As with any new technology, especially one that’s easy to use, there will be instances of people digitally modelling everything they see, so preoccupied with whether or not they could that they didn’t stop to think if they should, or need to. That’s fine… it doesn’t really advance science, and it can be frustrating to see instances of people doing something because they can, rather than because they should, but it doesn’t really hurt if people make the models or photos available in the long run, it’s all just more data for people to use. I guess the harm comes from people who are unfamiliar with the technology only ever seeing it as a means to make pretty pictures, rather than a means with which to genuinely enhance scientific research, communication, and documentation.

The impetus, then, is on the scientists who do use digitization regularly to make use of it as much as possible, and to highlight what it brings to the science when they do (I’ll have a blog post in coming weeks on why it’s so important for documenting tracks). Then, make sure people know about the guides that are freely available. A recent paper (Mallison and Wings 2014) in Journal of Palaeontological techniques goes into more detail specifically for palaeontologists, covering the scenarios that can be difficult.

I genuinely find myself amazed by the methods available to palaeontologists today, it’s nothing short of magical. Being able to head into the field with a small device in your pocket (that’s also a communicator device, a personal assistant, or just a camera), which can take a few photos and turn them into a digital 3D copy of the object, specimen, or entire outcrop before you… It was verging on science fiction when I started my PhD, now it’s a core part of my research.