CEO at age 24! What a great way to splash into the CG World! Matt
Sederberg introduces a new and innovative company called T-Splines,
LLC in Orem, Utah, nestled in a valley of the Rocky Mountains. Matt
studied Economics at Brigham Young University and he spent his last
two years of college participating in business school contests,
picking the brains of entrepreneur mentors, and mixing with members
of BYUs award-winning animation department. His goal was to go
into business in new technology. That goal is now realized in
Matt's company that produces cutting-edge technology. With the
company T-Splines, the future of 3D modeling is here. Computer
Science professor Tom Sederberg, Matt's father, is the one who
laboriously perfected the T-splines formulas over two decades.
This new formula revitalizes NURBS and brings new control to
modeling. Complex as this sounds, the average modeler can use
T-splines technology. So how do you really get to be CEO at age 24?
It may be a combination of hard work, independence, dreams, a
vision, and the magic of a product that needs to be shared with the
world. Matt says: My dad and I, weve always been quite close.
Hes always had a very high standard of his academic research, but
while I admired what he did and even was attracted to mathematics
and computers, where his talents lay, I had an independent streak
in me that resisted following in his career footsteps: I wanted to
do my own thing. As I progressed through college and became
fascinated with how the economy works and how businesses operate in
a capitalist society, I was eager to begin to apply all my book
learning about supply and demand curves from economics, and
actually get behind the controls of a business myself. I never
would have guessed that my chance to start and run a business would
come so soon before I even got my diploma. In 2003, Kyle Welch,
a lead from BYUs highly-regarded Entrepreneurship program stumbled
across some projects my dad was incubating in his research lab, and
convinced him to enter some of these inventions into the schools
prestigious Business Plan Competition. Kyle contacted me, and while
the two of us fine-tuned a commercialization plan for T-Splines, I
also got a generous amount of mentoring of how to run a business
from BYUs resident entrepreneurs, and local industry folks, who
were intrigued by the T-Splines idea. We ended up taking second
place in this competition and earning $17,500 in cash and in-kind
servicesenough that my dad and I decided to go ahead and
incorporate the business in late 2004. The most rewarding aspect
of my job has been being able to help my dad commercialize the
culminating invention of his years of research. We both view it as
fortunate timing that his discovery and my coming of age aligned so
well. It has also been rewarding for me to be able to make my own
unique contribution to our company: while my dad and our
researchers continue to develop new applications of the T-Spline
technology, I have been able to head up the business development,
which is where my own interests chiefly lie. Finally, my dad and I
have enjoyed being able to create jobs and establish a company
culture that reflects the values that we both share: hiring the
smartest minds available, working extremely hard, and being
supportive of the family responsibilities of our employees.
Now, who and what are T-splines? The company T-Splines has
developed a new way to manipulate NURBS control points and the
resultant mesh. The technology T-splines are a superset of
NURBS, which contain all the data that a NURBS model contains and
more and less. T-splines technology is a tool for artists to
model in ways that were not mathematically possible before. The
buzz phrase is "local refinement of surfaces." NURBS stands for
Non-Uniform Rational Basic Spline (pioneered by French engineers
Monsieur Pierre Bezier of Renault, and Monsieur Paul de Casteljau
of Peugeut). In short, T-splines technology is a method of
simplifying and reducing the number of control points in a NURBS
model, as well as adding control points where needed, while adding
to the high detail necessary for complex shapes. T-splines, is also
a subdivision tool. T-splines, is not just a polygon reduction
tool, even though it has the ability to reduce polygon count. With
T-splines, less data equals faster modeling, easier editing, faster
rendering and reduced file size. Simply put, T-splines, is a
revolutionary new way of organizing precision control points.
Partially funded by the National Science Foundation, Dr. Tom
Sederberg also created a new "surface intersector algorithm based
on T-splines." This means that there are a few new tools that the
NURBS modeler can use to produce more organic shapes, more precise
shapes, and do it all with far fewer control points. In Dr.
Sederberg's paper, T-spline Simplification and Local
Refinement," he states, "A serious weakness with NURS modeling is
that NURBS control points must lie topologically in a rectangular
grid. This means that typically, a large number of NURBS control
points serve no purpose other than to satisfy topological
constraints. They carry no significant geometric information.
T-splines are generalizations of NURBS surfaces that are capable of
significantly reducing the number of superfluous control points
from the NURBS model. The main differences between a T-mesh (i.e.,
a T-spline control mesh) and a NURBS control mesh is that T-splines
allow a row of control points to terminate. The final control point
in a partial row is called a T-junction."
Perhaps the easiest practical way to understand the difference
between T-splines vs. NURBS, is the visualization of curves vs.
flat polygonal mesh. In a flat mesh, to make a curve, one has to
place a great many control points along the path that the curve is
to be formed on. A sphere, for example, would require an infinite
number of points to result in a spherical looking object. Infinite
simply is not possible. Therefore, a finite number of points to
make a sphere look like a sphere results in a high polygon count
and at render time, still shows the flat surfaces and a rough
looking model. Obviously, the lower the count, the rougher the
resultant model appears when rendered. Render engines are capable
of additional smoothing according to the data in the model. Thus a
NURBS model which has the ability to generate curvature information
between two points (no longer flat mesh) is able to create a sphere
with far less points than is necessary in a flat polygon model.
Unfortunately, the curvature information in NURBS is limited to
averaging the curve between points. Some technology allows for
cages to be built and curvatures defined by the cage. While this
does result in a smooth sphere it still takes a substantial number
of points to define what that sphere is to look like. In more
complex models such as a human face, the number of control points
necessary for compound curves increases dramatically. Thus,
T-spline technology has the next level of curve interpretation. The
algorithms in T-splines formulate by Dr. Sederberg are such that a
curve can be interpreted outside of the average between control
points, or outside of the average in a cage. Curves can be
interpreted in weighted and more organic forms, and as such,
require even fewer control points to produce the same and modified
smooth flow of a NURBS model. T-Splines will be particularly
powerful for CAD applications because, unlike NURBS, T-Splines can
represent an entire CAD model without any gaps. This capability is
likewise valuable in animation with applications like Maya. For
example, a NURBS model that is stitched together with various NURBS
patches, such as one patch of neck with upper torso, and another
that does the head and face, is subject to stitch separation in
extreme movement positions in animation. Not so with T-splines. A
NURBS model converted to a T-splines model will not break apart at
the stitch locations. The addition of control points does not
change the surface. Additionally, the reduction in control points
makes the model faster to manipulate and animate. Thus, the ability
to add where needed, and reduce where not needed, is of great value
to modeling efficiency.
Another advantage of T-splines is to revitalize the older NURBS
models. Many modelers have scrapped their old models because they
were made in older NURBS programs, and the time to upgrade the
model is longer than it might be to just build the model from
scratch in modern software. By way of example, the Zygote Media
Group has made their 10-year-old male and female models available
as free downloads on the T-Splines website. These downloads are the T-spline
converted versions so that modelers can explore and compare the
differences.
This is where T-Splines shines for companies like Zygote that want
to resurrect older models with minimal time expenditures. An older
model that might have a huge file count, or has stitching failures,
or that simply needs some detail updates in facial features when
converted to a T-splines model is once again useable at modern
technology levels. You never know, we may one day see the Dork back
in action again.
The future of T-Splines is a wide-open field just waiting to be
developed. One of the focuses on the future is meeting the needs of
organic NURBS modelers. Certainly the gaming industry will look
long and hard at this Maya plug-in as a way to bring greater detail
over limited bandwidth. T-Splines plans to continue with aggressive
updates, and is currently only available for Maya. T-Splines can be
purchased through the T-Splines store. It requires Maya and will not
work with Maya PLE . 
- T-Splines Full Version (with CD): $799.00 [USD]
- Downloadable Version (without CD): 791.00 [USD]
- Academic Version: $199.00 [USD]
- Learning Edition (save disabled) is available for free.
 Tech Views is a regular featured column
with Renderosity Staff Writer/Sr. Tech Editor Eric Post [EricofSD].
December 27, 2005 |
Comments