david dvorak by lindash


									David Dvorak
CIS 736
Paper Review

                            “Organic Modeling Paper Review”

The first paper I read covered anatomically based modeling of muscles, bone structure,
and skin. The paper describes software that enables a user to specify skeletal structure,
insert muscle mass, and then cover the body in skin. After the model is set up the user
gets an accurate view of how the skin changes appearance with muscle movement. This
paper’s intended audience must be the upper echelon of computer scientist, because it
was very thick in rhetoric and very difficult to read. It was very interesting, but it was
still difficult to understand because they failed to explain complicated subject matter or
programming techniques. I would contend that this article would only be of interest to
other developers involved with similar anatomical modeling. It might go over a
biologist’s head if they didn’t have the proper computer or mathematical background, so
although it would seem relevant to a biologist, it might not be. According to the article,
no one has ever tried to model individual muscles, and the accuracy with which this
system attempts to model muscle movement is amazing. Their method for representing
the muscles with deformed cylinders seems very sound and thorough. I believe that their
system is able to achieve a significant degree of anatomical accuracy, and it’s no doubt
the most accurate system that has been developed so far. The paper is certainly
significant as this anatomically based model certainly has a wide range of potential
applications in multiple fields of study. However, it does have certain deficiencies. The
paper mentioned that the model groups some muscles, divides a few, and ignores the
others. Although the authors briefly state that the ignored muscles would have a
negligible impact on the model, they never exactly explain why. Also, the model is set
up so that the end slices of muscle do not move, which I believe poses a significant threat
to the realism of their model. I also wonder if they model the elasticity of skin
appropriately. They state that the skin has a stiffness value at rest and then can be
stretched during muscle motion. However, I wonder if the system is able to handle
instances of sagging in skin. Not all animals have skin that clings tightly to the muscles,
many have sagging patches of flesh. It seems that ignoring this results in an idealized
model, which would certainly give up a great degree of accuracy and realism. Nowhere
does the paper mention the ability to have skin that has the potential for the spring
stiffness to be lower than the spring stiffness of skin at rest. Finally, although the pictures
produced by the model are very impressive, they seem to be lacking something. I’m sure
that they are computationally accurate, they just don’t look realistic. The muscles look
like random bumps on the body. In my experience normal muscles that haven’t been
built up or toned do not produce any noticeable deformation in the skin, yet the pictures
seem to show every single detail of the muscle beneath, which is contrary to what you
observe in real life. The article suggests increasing skin thickness to model plump
animals, but that seems horribly inaccurate. In the end, the model seems very impressive,
yet the paper is thick with rhetoric and probably would only interest computer scientists
involved with similar work
The second paper I read discussed how to model large numbers of plants in a simulated
ecosystem. I found this paper very interesting. It was very easy to read and understand, I
think anyone could read it and be impressed by the magnitude of the modeling being
discussed. As stated in the paper, modeling plants in an ecosystem can have many uses.
Landscapers could use it to visualize what their plans look like, ecologists could visualize
the results of cutting down a particular section of forest, or a gardener could use it to
simply figure out how they want to plant their flowers. A very diverse audience could
appreciate this paper and it seems to have no particular group in mind to target. I believe
that this paper is very significant because it has a wide range of uses that can be
appreciated and applied by a variety of people. Their system for modeling the plant
ecosystems seems very good. I like their tool for specifying population density. As they
explain it, you can use a paint program to position dots that represent plants in the
ecosystem. This seems very simply, although I have to wonder how easy it makes it for
the user to visualize what the plants will look like. I don’t know how sophisticated of a
user you have to be before you’re able to easily visualize what the ecosystem would look
like based on the dots you’ve drawn. Furthermore, the density logic seems
oversimplified. The paper mentions that if two dots overlap, the bigger dot will win the
competition for resources with the smaller one, killing the smaller dot. I could
understand this happening in an already dense population, but what if there two flowers
that are the only flora within a fairly large area. If a smaller one overlaps with a larger
flower, it would die in the simulation. However, I do not believe the smaller one would
die in real life in this situation. I like their methods of rendering. One method they use is
to split the whole scene into sub scenes and then put them all together later. This works
great for big scenes but prevents global illumination. Their other method is using a ray-
tracing program called ray shade. Ray shade seems to work well for smaller scenes, and
its use of instancing is significant. The other ray shader they use is Toro. Toro is used
when the scene is too large to fit into memory. I was disappointed they didn’t go into
more depth about the capabilities of Toro. The pictures produced by their plant
ecosystem modeler are amazing. The look very realistic and are extremely detailed. The
best pictures are smaller scale. The larger rendered seems don’t seem as realistic.
Specifically, figure 15 doesn’t seem very realistic, mainly because you can’t see the
ground. I’m not sure if I’ve ever seen any instance in nature where flora is so dense that
you can’t see the ground. Figure 7 is much more impressive because you can see the soil
beneath the flowers. But figure 15 just looks like a chaotic smorgasbord of random
plants forced to live in an ecosystem together. It’s hard to believe that grass could reach
the branches of the trees in this scene. But I would say any faults in figure 15 are a result
of bad density specification and poor positioning within the ecosystem, and not the result
of poor rendering. The rendering capabilities of their system are incredible. The paper is
technically sound and would appeal to anyone. It is easy to read and would amaze
anyone interested in computer graphics and realistic modeling.

To top