Capturing a Dinosaur: Using Digital Tools to Reproduce a Physical Model

When I first visited the library at the AMNH, I was amazed by all of the physical items in the collection. As a public library, I expected to find books and publications on file, with the special Natural History and AMNH twist thrown in – original wonder cabinet catalogs, written log notes of old explorers, and such. But I was unprepared for the incredible art and memorabilia from across the history of the institution, and its potential for digitization.

One of the first items to catch my eye was a nearly 100-year old concrete model of a Camarasaurus (the associated signage reads “Erwin S. Christman (1885-1921)”). With permission from the library staff, Ariam, Nathan and I decided last week to test our mettle: had we learned enough since January to capture the dinosaur, refine the digital model, and print it out, all in just a few hours?

In the spirit of full honesty, I have to admit this was our second attempt. We had tried before, a month or so earlier, and resulted in not one successful capture. Using our trusty iPhones and 123D Catch, the capturing felt right but always turned out wrong.

This time, however, we could learn from our past mistakes. The first thing you might notice is the newspapers. Having lots of constant visual noise around the subject turns out to be much better than the empty reflective solid space around the base. The second thing I changed was the angle of my photos. This model is too long to photograph in one shot – I thought  I had to keep walking around the long side, my camera lens parallel to the dinosaur; I’ve now learned that to capture the long parts I should stand in the middle and rotate to capture the string of photos which, stitched together, will define the length. Finally, I changed the order of the photos: I took all of the photos first in a horizontal circle around the full model, then followed that up with a vertical circle as well.

We did it twice, and only one of the captures worked, but it worked perfectly. With almost no need for corrections, it came out like this:

The problem, we soon realized, was the head. It has an overhang – since it dips down there is nothing to start building it from. This was our first time dealing with an overhang. To date everything we’ve printed (designed by others) was ready to roll. First we tried a raft, which adds a layer on the bottom (not only unnecessary, as it already has a base, but unhelpful, as that didn’t reach the head). Then we turned on supports. This was interesting. Around the model the printer laid a support system to encapsulate it, designed to be removed once completed. The model printed great – probably the best of any attempt – but it was impossible to remove the supports, perhaps because they were printed right into the base.

In the end – and I am not totally happy with the solution – we redesigned the digital model and pulled down a line from the chin and dragged it to the ground. This pillar could offer the head the support required:

Btw, here’s a Vine I made capturing the reflection of the overhead lights on the wall shining through the model as it was being printed:

Once printed, the pillar can be removed. It’s not perfect, but feels good enough. Good enough, at least, to share back with the library staff. Here they are getting acquainted with their new dinosaur:

Our next step is to do some due diligence on this model. Anyone at the museum who looks at it can tell right away it might be historically accurate – it’s what we believed about dinosaurs at the time it was built – but is nonetheless incorrect. For example, dinosaurs did not drag their tails around as this one depicts. So we need to collect some contextualizing and meta data to associate with this model.

The second step is to build an official AMNH-branded and approved base to rest this on top of. Once we do, perhaps it will be able to join its younger sibling – one visual representation next to another: