Texture Pipeline

Textures in the 3D space will be broken up into 3 distinct materials. Initially, these materials are represented by three separate colors: red, green, and blue. Each material will include a layer of painted detail in addition to the solid colors. These details will maintain their spacial relationship to the shapes they are applied to. The solid colors will eventually be removed and replaced with flat patterns and textures that will appear to be behind each object. To help distinguish objects from one another, the darkness and lightness that naturally occurs on each facet of a 3D object with a light source will be maintained.

Nuke pipeline:::
My first attempt at swapping out colors involved using a chroma key to pull out a specific range of color from images rendered out of Maya. This process was not as accurate as I needed it to be, and the result varied from image to image. I needed a pipeline that would work in all cases.

Next I tried limiting myself to three colors, RGB, and moving image channels around, which turned out to be faster, cleaner, and worked well enough for multiple cases. The unfortunate downside is that each image sequence out of Maya would be limited by these three colors only, and I would be forced to render out multiple passes from maya to give myself the texture diversity I want in the final comp. For example, the main character requires three colors alone, and does not share any textures with objects in the environment. He would be one pass out of Maya, each boxcar type would be another pass, the ground and railroad tracks might be another, etc.

Here is an example of this effect currently used today.
http://www.youtube.com/watch?v=rWtecq1hRDw&feature=related

Story Change

I changed the story for a few reasons. One of which was because I felt
as though the abstracted visuals I plan on taking advantage of are
better suited for a storyline that takes place in a fictional world.

Swapping between the abstract view and the normal perspective view is
no longer a factor in this animation. One focus of this thesis is
placing an importance on the bond between visual abstraction and
storytelling. The 'look' of this final piece should not be treated as
a replaceable component. Instead, the story is relying on particular
elements to be depicted in certain ways. With this in mind, I felt
that having part of the story depicted in a more traditional manner
did not add anything substantial to either the story in its entirety,
nor to this thesis. When viewed as a whole, the concept for the
original story also seemed to emphasize a positive bias towards the
normal space perspective, such that it acted as the 'reward' for the
main character. While indeed, elements are to be composed abstractly
for the audience's benefit, I decided it would be better if characters
in the story remained unaware of the visual abstraction.

I thought this was interesting

The game uses eye tracking to determine where you are in relation to
the physical screen, and then compensates by adjusting the virtual
camera.

Rigging

The standard rig is controlling null objects which are above the
geometry itself, allowing me to manipulate the geometry while
maintaining joint based control over the character.
Problem areas include:
- Shoulders/clavicles are non-human, I had to use Set Driven Keys* to
constrain movement.
Animation changes - After doing test animations, I realized I needed
additional rig controls to avoid geometry clipping, as well as to
grant the character a wider range of movement.
- eye needed to move forward and back
- Added a few fail-safe controls to avoid clipping. These were
controls that would allow me to freely move entire components of the
body separate from the joint based rig.
Clusters have to be set to relative, otherwise the lattice points they
control will receive double transforms

*at its lowest level, set driven keys are used to form relationships
between two or more attribute variables.

Modeling

Translating the character from 2D to 3D was made simpler by the flat
shapes in the design. The challenge came when making sure each
component was able to shift and rotate appropriately in relation to
surrounding components.

At first, the eye, chest sphere, and hip joints were built as convex
3D objects. This produced a spherical shading gradient, which I
decided to be undesirable for this project. Instead, I opted for flat
circles that would eventually aim at the camera, preserving the
illusion of a sphere while eliminating the gradient.

The red chest "orb" was moved behind the 'rib' elements, to make it
less prominent. I did not want this object to be as important as the
eye in the animation.

Other examples of where I strayed from the concept image include: I
removed the slits in the chest, added single hinge joint objects to
the elbows, and added a wrist object to place the fingers on.

I decided to model using quads only, but not concern myself over
modeling for sub-divisible surfaces, especially now that I do not plan
on switching between the cubist view and normal perspective in the
animation. Intersecting objects (links between shoulder and forearm)
are acceptable. Hard edges define the boundaries of each component
regardless of whether or not the object is a single mesh.

Character Concept 02

Character Desgin

I wanted to make the lead character belong in this expansive field of
boxcars, so I vaguely based him on the look of older locomotives. This
is especially clear in the exaggerated chest and sternum.

I used Adobe Photoshop to create this concept image. To best represent
the final vision of the project, I had to think geometrically,
restricting curved edges to a very select number of components
including the eye, center 'orb', and small rivets.

The collection of color shapes was interesting, but to add a level of
cohesiveness I overlayed details that permeated multiple components.
Doing so really brought everything together, and was the inspiration
for separating base colors and detail textures when shading the 3D
elements in the final composition.

Once the body was completed, I felt that giving him some sort of staff
or pole to hold would better sell his purpose to an audience. The
staff suggests a level of authority, and immediately says to me (when
held in a specific way) that the lead character is guarding something,
or blocking the way. My goal was to make it apparent that this
character was either a shepherd or scarecrow for the trains he
oversees.

Character Concept

Above is a concept for the lead character. Currently I am working on translating him to 3D (progress below). Part of that process involves selecting key components that I will later add additional controls for the camera influenced skewing. I've decided to model using quads only, but not concern myself over modeling for sub-divisible surfaces, especially now that I do not plan on switching between the cubist view and normal perspective in the animation.