The Ongoing Trilobite Reconstruction Project Page 1 2 3 4
A short excerpt from about three minutes of video that I've shot of a fairy shrimp swimming in a petri dish on a microscope stage illuminated by cool fiber-optic lamps. In it, you can see the apparent physical similarities between this fairy shrimp and Opipeuter, down to the eleven pair of legs and heads with large, trailing protrusions. This similar anatomy is what, I presume, leads paleontologists to guess that Opipeuter may have swum in a ventral-side-up manner, as illustrated here by the shrimp. I was thrilled with the clarity of video I was able to get with a dark-stage setup, although this will be used more for motion study than anatomy. After raising this shrimp and worrying over it for days, I'm not about to dissect it, so I've ordered a batch of ten preserved specimens from a biological supply house which I will be dissecting and photographing under a microscope for anatomical details of the limbs. It's the equivalent of calling a hit on some shrimp, I suppose. But they're not my shrimp.
Please click on the image for video. Video is approximately 350K, 640 X 480, .mov format, and will open in a new window.
August 28, 2004
After almost a year of on-and-off research, reading the relevant sections of Treatise on Invertebrate Paleontology, by Moore, et al., The Appendages, Anatomy and Relationships of Trilobites by Raymond, and A Biology of Crustacea, by Green (for the description of feeding methods and swimming patterns of extant crustacea), as well as e-mail conversations with the abovementioned (and extremely patient and tolerant) Sam Gon, among others, I have settled upon the configuration shown at the left as the best appearance for Opipeuter's limbs. I strongly emphasize that any errors in this concept are my own, and not due to any input from Dr. Gon or any other of the people who have been so kind in their guidance. Since no known limbs of Opipeuter have been preserved, this is entirely speculative, and based on no hard data. These comments are meant to clarify my guesses as regards these limbs, not to claim accuracy or authenticity where there is none. Like all known trilobite limbs, they are biramous (having two forks), where the upper fork is the called the pre-epipodite and serves as both gill and paddle. The lower fork, the telopodite, is the walking leg. In all known species where the legs have been preserved, it is the pre-epipodite which has shown the greatest adaptation to the lifestyle of the trilobite. The telopodite remains standardised, consisting of a coxa, or basal segment, which may or may not be used in grinding food and moving it to the mouth, followed by six more-or-less similar segments . I have adhered to this plan for the telopodite of Opipeuter, although I have made it slight of stature on account of its presumed minimal utility in the pelagic animal.

I have based the overall structure on the limb of Kootenia dawsoni, as represented in the Treatise of Invertebrate Paleontology. Kootenia and Opipeuter were not closely related, insofar as I know, but the limb shape: paddle-like with slight signs of transverse segmentation, is very similar to the limb design of fairy shrimp as seen at left. The slight transverse segmentation may be indicative of points of flexion, as in the fairy shrimp limb, allowing the trilobite leg to bend on its recoil stroke, which would allow for a very powerful propulsive effect on the extended forward stroke. I'm choosing to interpret it that way.

In the photos on the left, you can see a frontal view of a pair of fairy shrimp limbs that have been cut out of the center of a preserved fairy shrimp. Of particular interest are all the lobes and paddle surfaces, as well as the fringes. In the case of fairy shrimp, the fringes are mostly inert fibrous extensions of the exoskeleton. In trilobites, the fringe-like arrays are actually gills. The gills are more like the slats of a venetian blind and may well serve as an extension of the paddle surface. It is presumed that the Trilobite's limbs' natural position is similar to that shown in the fairy shrimp to the left: with a backward curvature allowed by the flexion of the exoskeleton. On the back, or power portion, of the stroke, the limb would straighten as seen in the above movie, with the wave of activity passing smoothly down the segments of the body and providing a relatively continuous thrust.
September 10, 2004
This is the model that I hope to use for the legs in the final animation. It was made by constructing a simple geometric polygonal model in Maya and then exporting it as an .OBJ-format object into ZBrush. The image below is of this base mesh. Once it was exported, I was able to use the amazing ZBrush toolset to add detail and sculpt the final geometry. I'll also be using ZBrush to paint the color and specularity (shininess) maps for export back into Maya for rigging and animation.
September 12, 2004
This is my initial experiment with exporting the sculpted leg from ZBrush into Maya. It was done by exporting a simplified mesh from ZBrush, along with a black and white image that represents the sculpted details of the mesh. This was applied as both a displacement map and a bump map in Maya, to get the final result you see here. The polygonal base object was converted to a subdivision surface object in order to give the displacement map more points to work with. Now that I know how to do it, I'll have to go back to ZBrush and make some adaptations to the mesh and displacement in order to fix some minor difficulties that this transfer has made apparent.

To see an animated spin of the limb, click the image at left.

Approximately 1 MB, Quicktime, Sorenson 3 compression.
November 23, 2004
Grrrrrrrr.....The displacement and bump map idea proved really intractable, primarily because it was just not efficiently scaleable. The single leg above caused a pretty hefty load on my computer, and the idea of having 22 of them all thrashing about was just not practical. Instead, I've settled on an idea from game design of using Normal Maps. These are a refinement on the idea of bump maps which allow a low-polygon object to mimic very closely the appearance of a high-polygon object, with the only drawback being that they (like regular bump maps) do not actually affect geometry, so they can lead to odd effects in the shadows, and can reveal their lack of detail in the lack of edge resolution. They are based on the concept of "Normals", a device used in 3D to determine in what direction a particular polygon is facing. By applying a map that tracks the relative direction of the polygons on a high-polygon object onto the surface of a low-polygon object, they allow the low-polygon object to render as though it were much more complex. They render very rapidly and efficiently, and I think that when the legs are all in motion, with motion-blur and speedy lashing, the small infelicities will be invisible. I've also switched to using the 3D program Lightwave, having been completely baffled by Maya's excessively left-brained implementation of Normal mapping. Things are going quicker now......
To see the Normal-mapped leg rotate, click the above image.Quicktime, Sorensen 3 compression, approx. 1.4 MB.

As you can see in the above image, the effect is not quite as good as the displaced version, but still plenty good. The main area of concern is the lower edge, which displays the faceting of the low-poly base object. I believe this will be unnoticeable in the final animation. Additionally, it requires very high levels of anti-aliasing to keep the ridges from "crawling". The above animation was rendered with Lightwave's "Enhanced High" anti-aliasing setting.
November 30, 2004
Now that I have the leg built, I'm hoping, hoping, hoping that things will go more swiftly. This is a test of the setup that I'll be using to animate the legs. It uses a plugin for Lightwave called "Cyclist", which manipulates motions of objects based on the motion of other objects. In this case, I'm moving the legs through the motion of a null, or invisible place-holder object. Once I have the limbs set up properly, I'll be able to speed them up and slow them down at will, and they will always remain synchronized.

At the moment, there's a lot of crossover and intersection in the motions, but that's because this is just a "proof of concept" to show that the method may work, rather than a finished motion test.
Click on the above image for test animation of trilobite limb stand-in objects. Quicktime, Sorenson 3, Approx. 560 K
December 27, 2004
I've spent the past few weeks assigning UV maps to the various segments of the trilobite anatomy. UV maps are a way of applying images to 3D objects so as to minimize stretching and distortions of the textures. It's similar to the stretching that you see in flat world maps, which always show the arctic and antarctic as long strips across the bottom and top. But that's jsut the beginning. The objects used in 3D are rarely as simple as a globe. You have to figure out how to cut up the object and distort the elements so as to minimize any weirdnesses, or rather, since weirdness is inevitable, how to hide the weirdness in out of the way places on the model.

This model has been covered with a checkered image so as to reveal any potential distortions as distortions in the square pattern. In the case of the trilobite, that "gather" on the top of its head will be covered by another image.

Another aspect of UV mapping is that it provides you with a template that can be used to draw the various textures required to fully realize how light will interact with the model. For more information on UV mapping, check out Leigh van der Byl's tutorials, which have helped me a lot.
Page 3