david's really interesting pages | Just another WordPress site

3D skeletal reconstructions?

(or download turntable animation.)

Recent efforts in paleoart have both inspired and intimidated me. Inspired, because there’s so much going on at the moment (visit OpenDinoProject, ArtEvolved for good stuff and good starts for more links). And I’m intimidated because of the high level of work out there, combined with the high level of respect I have for the responsibilities involved in representing these exciting animals.

Case in point: skeletal reconstructions.
After facing the many issues involved in creating a Supersaurus sauropod or a monster newt (they’re both approximately the same amount of work, despite the difference in size and popularity) I began looking for ways to 1) increase the respect for research and 2) bait scientists with visualization (sic: art) that has the potential to enhance current methods of organizing and manipulating existing research. Both goals seemed to point to the skeletal reconstruction. Gregory Paul (I’ve become a huge fan of his writing and art) credits these for playing an important role  in the modernization of dinosaur art and theory. Here’s an example:

This is Tuojiangosaurus multispinus as reconstructed by the industrious and generous Scott Hartmann. (Thank you so much, Scott!) As you can imagine, this is an invaluable reference for the artist, particularly the non-scientist artist such as myself. But its also missing quite a bit of information. Some reconstructions include top and front/back views, but those are rare. So we’re often left with a 2 dimensional floorplan for a 3 dimensional house. So I thought “hey, wouldn’t a 3D reconstruction solve all this?” and set off to model the gerrothorax skeleton (the newt).
Result: the effort is unjustifiable (at least for the goal of illustration). Perhaps scanned meshes will change this, just as museum mounts use molds of the real bone. But I don’t think so. The first warning I encountered when researching my interest in paleoart was “beware the museum mount”. Its likely incorrect.

(No reference to these specific models, illustration purposes only. Sources: a and b.)
And worse: incorrect or not, it can’t be perfect. Visually, it claims truth via realism while its content lags behind this claim. There’s an interesting distinction between these displays and the graphic quality of the skeletal reconstruction above: abstraction. Abstraction is basically a selective reduction of information. This reduction is inherently broadcast within the visual style so that the viewer immediately perceives the representation as being abstracted. Scott’s drawing immediately reads as a representation and is accepted for the information it conveys, while the pterodactyl above reads as a realistic model, yet the viewer feels the discrepancy to an actual bone.

So I’m in conflict between the wish to raise accountability in my reconstructions without selling them visually with more authority than they deserve. Fortunately, the solution (or at least one such) lies smack in my area of expertise: stylization.

As a contract job had me working on a human character, and because human skeletons are comparatively well understood, I started with that. It took ~3 days to model, then I quickly lit and shaded, with a flesh volume blacked out as is custom with the graphic reconstructions. My hope is that no one will look at this and think “oh, that’s a skeleton.” They’ll recognize it as an illustration, yet still accept the information that the reconstruction is trying to convey. This will hopefully do the following:  One, it will look cooler. I hate those bone shaded 3D skeletons that are neither here nor there. Two, it’ll allow for detailed communication of things like biomechanics or evolution. And third, it will hopefully allow a different way of perceiving the bone measurements and reconstructions and thus serve gains in scientific understanding. Time will tell.

What’s going on is that the bones have reduced detail. This reduction isn’t linear, however. The rotation points and muscle attachments are alluded to and partially remain quite sharp. The rib cage is more or less a volume with planks placed on top (they can be removed for less detail). The “industrial” shader is a ramp from warm to cold based on incidence angle to the light source. The “pearl” shader is a ramp from white to black based on incidence to camera. These serve the function of making the volume of the bone forms themselves more legible.  The idea is that the bones remain crisp and aesthetically pleasing yet clearly show that they do not physically correspond to the actual bone. Result: (I hope) function and morphology are communicated, without the pretense of having a true bone reconstruction.

Feedback is very welcome!

There are 2 Comments to this article

drip | david’s really interesting pages… says:

[...] there are also some pitfalls. As I mentioned earlier, a swell-looking skeleton can allude to accuracy that it doesn’t possess. This is comparable [...]

drip | david’s really interesting pages… says:

[...] talked before about my distaste for 3D skeletons that visually purport to be real, when they in fact are inexact [...]

Write a Comment