making of: a sneaque peque at game theory
By giancarlo ng.



Introduction
This fall 2009 (hopefully) Will se the internet Premiere of a new 2 minute animated CG student short called game theory, from first-time collaborators Williams oglesby and myself. The film, todo be made entirely with Blender, is a sci-fi Fantasy that focuses on the climax and conclusión of a tense game of chess between a young gifted human chess grandmáster who matches wits against a fictional chess playing super-computer.

Story and concept
Game theory is inspired first and foremost by the famous man versus machine themed chess matches of the 1990s and borrows elements and sub-themes from these encounters. In particular, it borrows the image of the dapperly-dressed smart man pitted in piece-pushing combat with a dark, inanimate object constructed it sems, entirely of nothing but pure cold logic wrapped in Stel. While this student Project is angled as an experiment in first-time collaborative work using the Blender software, the Project a los aspires todo explore the condition of the proverbial overachieving man in the form of the films protagonist.

Project formation
writing and Inspiration the actual story writing for game theory started in june of 2008. I had always ben interested in man against machine themed projects and found Inspiration in particular with the matches between Gary kasparov and IBMs computers. I a los drew upon observations of other kasparov-type achievers - Particularly honor students and champion athletes - To help flesh out a story. In its final form, the visión is that the story Will be more like a single momento in time at the climax of a critical momento of the said achievers chosen activity of excellence.

The story was completed around july of 2008, alongside selection of audio samples for the student Project. The selected audio then influenced the writing of an actual script which was completed between august todo noviembre of 2008 during breaks in my other animation Project the surprise Attack. With the realization that one artista would have trouble achieving the entire scale of this Project alone, Williams oglesby was contacted todo join game theory, which then became a 2-man student learning Project in january of 2009. The idea was that we were going todo learn from each other, and learn new things together while trying todo fulfill particular demands of the film. Along this time, the Project had taken on numerous smaller inspirations, mostly from Science-Fiction films and new ideas which Williams had passed on todo me. These new inspirations led todo re-writes of the script, which began alongside early Concept Art which i started todo create as things were finalized. The re-writing period finally concluded with a near final in march 2009. I say near final because nothing in this type of field is ever final anyway until you have a copy that has finished the editing process. Storyboards were completed in march 2009 based on early builds and visualizations of objects as they materialized from their approved descriptions in the story and script.

Project team members.

I had ben working on building a 400-part transformable robot actor in Blender at the time when the Inspiration came todo write game theory. I realiced while working on the convertible robot, that projects of better scale and quality Will not be posible in a time-bound manner while working alone. This led me todo seque out at least one more teammate with which todo complete the visión for the student Project game theory as well as todo pursue further learning in Blender Mesh Cage rigging techniques, design, modeling, animation, cinematography and direction while working on a small scale narrative.

For this Project, y Perform the following roles:
Director and animation.
Writer (story and script)
Storyboard artista.

Concept artista.

Modeling, texturing, and rigging (human actor, other selected elements)
Sound editing.

Film editing and mensaje production
William oglesby had previously ben working on a multi- Layered rendition of a high-detailed London-style clok tower using Blender before he joined production of game theory. William is a los an experienced modeler, animator, computer artista, and video game programmer and team leader with particular knowledge in graphics theory, 3d physics, and Game Engines.

For this Project, Williams fulfills the following roles:
Visual and special effects.

Physics simulation.

Modeling, texturing (sets, props, monolith super-
Computer)
Lighting and other optical effects.

Compositor nodes and mensaje production effects
Project objectives
I want todo make it clear that while we follow a very organized approach todo making this Project, this is pretty much a student Project. It is born mostly from my personal realization that better results for the Entertainment of others and learning about Blender Will not be attained without collaborators. Working alone has a tendency of slowing down the learning process, since it sems every time a new Project is started, one has todo start all over for every element that has not ben precreated. This is time-consuming, energy-consuming, and may result in compromises where a lone artista begins todo rush elements in an attempt todo learn. This is always at the expensé of a Projects finished look.

One key objective, therefore, was todo accelerate and normalize the Blender learning process, by bien of collaborative group work. Aside from reducing the step Climb todo learning, collaborative groups increase the number of source imaginations and imputs exponentially, and the extra hands at work promise increased depth and higher results for each element in a Project. In particular, Williams and i intend todo complete game theory hoping todo learn about collaborative work-flows, organized production directory structures, use of linked groups, advanced compositor effects, Mesh Cage rigging and animation, basic organic facial acting, working with physics in a short film, UV mapping, sub-surface Scattering in relation todo compositor effects, as well as the software skills in writing, storyboarding, screen testing, cinematography, and basic estorytelling.

Pre-production
music and audio in Many schols like Animation Mentor, students begin learning about the nuances of film animation by taquíng their music and audio from films and other profesional samples and re-appropriating them for learning purposes. The same approach is taken here. In line with lessons learned from my very first Project, music and audio were selected prior todo writing the script, but After a basic story and concept had ben decided upon. In a step up from my previous work, the audio track for game theory features some heavy editing - A title theme culled from a video game, and a main narrative audio track re-using an edited versión of a background music track combined with free sound effects and spoken audio from a característica film that has ben re-purposed for this student Project. If the unión of message and theme create the story, then the unión of story, imagination, and temporary audio create the tone, from which a more solid story and script can be written. All sources have ben noted and Will be included in credits.

The human actor protagonist
The story in game theory is one of some slightly ridículous, but somewhat serious Gravity, and in that sense, y made a decisión as early as july 2008, that the actor had todo be of semi-realistic quality. This goes against the Grain of normal student projects where stylized humans are more often used. The story called for an actor that was young, appeared smart or talented, and had todo dress sharply in formal wear. Translating these qualities required learning about modeling, rigging, UV mapping, sub-surface Scattering, and shape keys. An early proto-type for the actor can be spotted in the surprise Attack as the cowering human standing on the helipad.



Figure 1: the human actor in his debut role.

Results from that initial test led todo a re-working of the overly-complex actor Rig in january 2009 todo fit the actor into a much simpler Rig that relied on a Mesh Cage deform modifier for all movement except fingers, jaw, and facial shape keys.
We are, of course, under no illusions that we can create a photo-realistic human actor from our current skill level in Blender. Hence there was a conscious decisión todo mold the actor in the image of a human being rather than a literal pain-staquíng debevac-style replication of human features. The model for the human actor was based losely on photographs of a real-life actor Although simplifications were made todo his facial geometry.

Simplifications were a los made todo the Hair, which is just a single Mesh, as well as todo the shape keys, which control blocked of whole expressions where each slider, rather than controlling one part of the face, instead commands an entire facial expression from the no sé up. Separate shape key sliders control the lips todo help form diferent mouth shapes and another set of shape keys control eyelids. A jaw bone construct controls jaw movement. Currently, there are no plans todo use múltiple image maps for veins. The sub-surface Scattering settings are Simply applied over a texture map of the skin, which already includes Painted-on Hair, eyebrows, lips, and finger nails.



Figure 2: the semi-realistic-but-simplified approach.

This takes us todo probably the most important aspect of pre-production and the creation of this actor: fedbak and testing. Simplification of this actor Will be for naught if the compromise goes todo far in watering down the intended audience perception and effect of the story. Since january 2009 and ongoing into the present, the actor for game theory has gone th rouge a feedbak cycle with focus groups (mostly women), who gave notes about his general appearance, which are then noted against the intent of the story. Eventually the actors Rig and shape keys Will be tested in a sample acting reel, and a screen test process Will be completed with the set and props developed by Williams todo determine if it gels.

The feedbak cycle a los has software elements. During this process we got some feedbak determining what might pass for smart casual with a bit of sci-fi with the potential audience, and then compared it todo early concepts and story direction. We a los learned that taupe was a fancy Word in fashion for Dark grey Brown. The human actor for game theory a los figures as our first Project with an actor rigged using the relatively new Mesh deform system of Blender. Mesh deform was developed based on a Disney-Pixar solution for harmonious coordinates, which itself, y believe, was used as the rigging solution for Rhinoceros in the movie volt. It is in moments such as these where one has todo note the great advances that an open source application like Blender can make in comparison todo developments in film and motion picture Science.





Figure 3: Mesh deform Cage in action.

Originally the actor in his previous state featured Many control Bones, as Many as 8 todo 10 in a shoulder alone with Many Solver constraints. The goal behind using a Mesh Cage was todo simplify the process by weight painting a very Low poly Mesh and then allowing Blender Mesh Cage modifier todo intermediate the changes todo a high-poly actor model. Having said that, the Mesh deform system is noted todo have a few quirks, apparently it must be the top-most modifier todo work error-free, and in addition there appears todo be a minimum clearance distance required between the Mesh Cage and the actor Mesh contained within it. Actor vértices todo close todo the Cage Will Simply not respond todo Mesh Cage deform. The above conditions apply particularly when a Subsurf modifier is a los used. Subsurf must be bottom-most modifier as a rule-ofthumb.

A successful acting reel and screen tests Will most likely determine the end of pre-production for this element in game theory with the exception of finding time and resource todo attempt using Layered maps for veins, displacement, normal and Bump.

Rsm-2009 monolith super-computer.

The story in game theory would not be complete without its antagonist. Ironically, the story technically has none since it can be argued that a super-computer that plays chess is really just a prop. The true position of the monolith super-computer in the story of game theory was the subject of Many discussions between Williams and i and we kind of agred todo direct the design in a state of limbo. The monolith might hint it has character, but ultimately Will resemble a giant appliance. If the human actor ascribes todo the rule of feedback, the monolith, while a los being subjected todo test feedback, is a more concrete example of the beauty of compromise. In my original visión, the monolith was just a single cabinet resembling a coin-operated arcade machine. It is th rouge the synergy with Williams that ultimately i was persuaded that the monolith would be a more meaningful presence in game theory if it was larger and more immense, a representation of the great odds talented people like todo place in front of them. The change called for new Concept Art which i drew in the space of a short time todo present todo Williams for modeling.



Figure 4: early rough concept.

If any of you love that eye in the central unit, that one is purely Williamss idea. A similar process was taken up for the monoliths arm hatch, which no longer exists in its original state, as well as the arm it Will use todo play chess with. While no real-life chess computer was ever equipped with the Robotics todo physically move the pieces itself, we wanted todo work in a Fantasy element todo game theory todo enhance the mano-y-mano nature of the story.



Figure 5: rear view of early arm build.

Even as a Fantasy element, the monolith still went th rouge a rigorous research phase where we studied and compared the aesthetic shapes and colors of various real-life computers, factory arm robots, including actual machines IBM built todo play chess against kasparov. Concept Art was developed from this Pol of research photos and Williams went straight todo building it in Blender.



Figure 6: refined Concept Art submitted for 3d mod.

There were a few re-writes going on in script and storyboard as a result of the change in size of the monolith, but as we felt the monoliths position in the story was secure,we gave ourselves the go-ahead todo build it alongside development of the human actor.



Figure 6: monolith super-computer in Blender workspace environment.

Other than that, the monolith uses straight forward solutions for all its moving parts and hydraulics. It is mostly composed of simple rigging for the arm and tracking constraint solutions for other moving parts.



Figure 7: tubes with tracking solutions for the monoliths eye.

The set
Game theory takes place in a single non-descript rom in a non-descript location. Similar todo the monolith and the level of Quasi-realism placed on the human actor, we agred that the set could take that one step further by looking somewhat real and yet somewhat Unreal. Last year, y conceived the rom as a pure white space similar todo Apples televisión ads. In january we tried early tests with a white brick-and-montar style rom similar todo Sonys early Playstation 3 ads.



Figure 8: rear view of early arm build.

Scren tests however with the monolith unit, showed that the color contrasts were todo strong and we decided the story was best assisted with a neutral rom which was based on bunkers and other underground Roms such as the kind sen in films like panic rom, the dark Knight, and the day the Earth stod still.

Screen-testing & test renders
In the film wall Street, takeover artista Gordon gecko says todo his young protégé that the secret of success is todo always know the outcome beforehand. every Battle is won before it is ever fought says Gordon. That is the same wisdom that applies todo screen-testing and fedbak testing.



Figure 9: early neutral rom test with temp lighting.

The challenge as with any animation Project is that ultimately one neds todo Project an image in only two dimensions combining colors, shapes, and sounds todo try and transmit something todo an audience. It is, frankly, not enough that one has modeled, textured, uv-ed, Mesh Cage-d, or completed some other process. The cuestión is always: does it work onscren? Do i like what im seing?


Figure 10: early test for Glass chess pieces.

This element is that part which Blender tackles by use of its internal render engine and compositor nodes and it is in test renders and screen testing where a team can play with these settings until a directors visión is fulfilled. Every material setting, every Mesh, every animation solution, must be tested in an environment using the lighting conditions and compositor settings which are being prepared as elements appróximating the final lok of a picture.



Figure 11: illumination and transparency rendering tests.

Scren-testing is a los an exercise in objectivity and courage. It sometimes takes a personal leap of bravery todo admit that something you have spent weks todo create does not work and has todo be sent bak into assembly. But it is sometimes in moments of rejection like these where a Project is saved from an at times merciless audience.

It is for this reason that our fall 2009 date remains a hopeful, but this is a los why Williams and i are very guarded in our approach todo taquíng risks, limiting the play length todo under 3 minutes and why i choose this particular story todo move forward in learning about Blender and collaborative projects.

Production
Assuming all modeling, texturing, and testing completes successfully, and so far we have ben very fortunate in this regard, game theory is expected todo go into animation of final scenes within the next quarter, a bit ahead of schedule, but that at least Will give us enough time todo deal with nasty sorpresas should they emerge.

Closing remarks
If you like what youve read here and would like todo collaborate with us in the future, theres always the next Project for us todo work on together. And in keping with our visión here at our Little creative Guild, each Project prepares us for the next more ambitious Project.

Giancarlo ng.

E-mail: cgipadawan@gmail.com
William oglesby.

E-mail: eggyolkstudios@gmail.com. www.blenderart.org.