Character Research

Improv: A System for Scripting Interactive Actors in Virtual Worlds – Ken Perlin / Athomas Goldberg

I appreciate Perlin’s and Goldberg’s user perception validation for Scripting Interactive Actors. Specifically, the Animation and Behavior engines allowed fine or broad strokes for the artist to define natural looking characters. The key understanding was that viewers do not perceive the mechanism itself, rather they perceive some presentation of the motion it produces (Perlin noise). The intention is not only a research endeavor but also on its application (and artist workflows) this area of development can foster. The Metaverse mentality was a little premature back then and perhaps more realizable today in Second Life, Klima’s “the music lounge”, and others that resonates with aspects of this research.

The EMOTE Model for Effort and Shape – Diane Chi, Monica Costa, Liwei Zhao, Norman Badler

There is a reason why agents look and behave artificially and this work clearly articulates the complexity of gesture alone. Using McNeil’s approach and Laban Movement Analysis as possible gestural grammars, it is quite possible to say that gesture and agent EMBODYMENT has its own expressive language layered on top of communicative speech acts. As with any language there is an inherent and agreed upon symbolic meaning these grammars represent as well as arbitrary or ambiguous meanings. This could occur especially in the “blended” or “serial” approach of gesture -what happens if an agent has an unrelenting nervous “tick” but is also trying to give detailed directions?
This paper seems to present intentional movements well but what about the unintentional?

For me this works resonates with previous AI work on believable and emotional characters – except embodied expression is highlighted. This paper does not seem to address physically based and more “ragdoll” (environmental) approaches which can also dictate movement above skeletal constraints. Lastly, A deconstructive authoring approach is necessary which is counter intuitive to traditional character animation techniques.

Lifelike Gesture Synthesis and Timing for Conversational Agents – Ipke Wachsmuth, Stefan Kopp

Continuing on with the notion of a language of embodiment, to what extent this language does and does not coincide with spoken language is another layer of subtlety. Kopp’s work specifically in hand gesture tied to specific temporal speech acts is one direct approach and commonality between languages, but what happens when there is incongruence in mapping? For example what happens if a character says she is listening when her body language is clearly saying she isn’t? Say one thing and do another is a common Murphy law. Facial expression is another example, someone can have an entirely demur expression when in fact she is being deceitful. There is clearly more work needed in this area and I can understand why the games industry is content to let animators emphasize scripting these complex behaviors.

From Linear to Interactive Animation: How Autonomous Characters Change the Process and Product of Animating – Bill Tomlinson

This work is broad but touches on issues discussed previously in the readings. The use of Alphawolf is an excellent example how collaborative teamwork and procedural approaches can create viscerally engaging emergent experiences. However the animalistic expressive nature of wolfs fails to address the nuanced nature and layering of meaning in gesture and language exemplified in human communication. Self indulgent we are as humans, is it no surprise the majority of our work is reflectively about us? Our heritage is ingrained millennia of theater, and more recently film and television portrayals of humanity. Can we really escape it? Tomlinson does provide insight into key differences in linear and interactive mediums of expression. The gem I found was in the importance of “technical artistry” in collaborative teams. The breadth and depth of desired character expression has to make technical trade-offs and creative-calls.

Architecture for Action, Emotion, and Social Behavior -Joseph Bates

Please see my final paper for the Directed Reading in interactive Narrative Technologies where I discuss the OZ project in further detail.

Anticipatory AI and Compelling Characters – Bruce Blumberg

You have to crawl before you learn to walk. This is not in any way diminishing the importance of “low level” anticipatory AI as the foundational starting point for more complex agent emotional and knowledge states. Let me repeat, Alphawolf is a fantastic example. I wonder how real dogs react to Alphawolf, this would be a great dog friend when you’re away at work by the way… Anyways connecting this research back into a humanistic domain seems like the logical next step which reveals new questions: how does spoken language reinforce a gestural language? How can we incorporate subtleties of character emotion, goals, moods, plans into compelling and seamless action?  How can we incorporate multiple layers of control (micro-meso-macro) characteristic of Perlin’s/Goldberg’s work?