Topher's Links: UCSC-Computational Cinematics Studio (CCS)... Contact: topherm@soe.ucsc.edu
|
Gestural drawings from A Dictionary of Theatre Anthropology: The Secret Art of the Performer by Barba and Savarese, 1991.
|
|||
|
CMPS203 Programming Languages: Survey of Functional Programming for Games (Faculty: Cormac Flanagan) by Topher Maraffi and David Seagal. My contribution is the second part of our FRP survey (starts slide #11 in the presentation, or Part 4 in the paper) , focusing on how the reactive media DSLs of Conal Elliott & Paul Hudak could impact furure NUI game development. From the Conclusion of our paper "Leveling Up: Could Functional Programming Be a Game Changer?"... "It is a possibility that FP has not taken off for game development because there wasn’t an ideal fit for it. However, the semantics of FRP, with continuous time and behavior/signals, seems particularly well suited for streaming Natural User Interface (NUI) applications that use Kinect or Wii technology. Could the paradigms of FRP and NUI be merged to provide a new kind of artist friendly tool that enhances content generation for games? The imperative method of keyframing a rig’s joints on a discrete timeline is closer to antiquated stop motion technique than to modern motion capture. FRP could fundamentally change the way animations are produced by virtualizing gesture into performance signal functions, which could then be composed directly through a NUI device like Kinect. This could fuse Hudak’s concept of virtual instruments with Elliott’s concept of tangible values in a new FRP approach for gestural performance in games."
CMPS262 Computer Animation: Engineering Procedural Mimicry of Gesture through Neuro-Evolutionary Performer Modeling (Faculty: Alex Pang) by Topher Maraffi and Dustin Escoffery. My concept for this project was to use the HyperNEAT neuro-evolutionary algorithm to implement simulated mimicry as performer modeling, recognition, prediction, and synthesis through a NUI interface device. This is still a work in progress...
CMPS240 Natural Language Processing: The Movienator! (Faculty: Marilyn Walker)
My concept for the Movienator! project was to design a task oriented agent modeled after the distinctive personality of Arnold, and which uses natural language processing (NLP) to help user’s get movie information from the IMDB database. The kicker is that The Movienator agent prefers Arnold movies over anything else, always trying to steer the user towards one of his many films, or towards a film starring someone he worked with, or at least a movie in the action genre. Filled with Arnold movie trivia and famous one-liners, The Movienator is meant to be entertaining as well as functional. I designed the new architecture to interface with a standard DM, NLU, NLG, and DB moviebot skeleton architecture from the previous year's class. I also developed the Arnold ontology as a list of tuples containing movie titles and quotes-taglines, with system preferences for genres and actors, and wrote getArnold and Bayes functions to interact with the Arnold knowledge base and to drive the system mood.
CMPS244 Game AI: Dance Mario Dance! (Faculty: Arnav Jhala) by Topher Maraffi (Project Management, Concept & Design, and Music Processing) and Ron Liu (Level Coding). My concept for the Dance Mario Dance! (DMD) project was to design a simple platformer game where the levels are dynamically generated from the in-game music, so that navigating the resulting level structures requires rhythmic gameplay, similar to Guitar Hero or DDR. Using Infinite Mario, we set out to generate level content (hills, tokens, etc) that would compel the player to make Mario dance through the level in real-time to the song. To complete the level successfully, a player should manipulate the game controls like they are playing an instrument, with difficulty levels being based on the musical complexity and tempo of the music. Our initial prototype in Java produced hills that reflected the shape of an MP3 waveform. I did most of the design, project management, and music processing research, while Ron did most of the level programming. Ideally, in future versions, I would like to generate level content from individual midi instruments, replace standard controls with a controller-less system like Microsoft Kinect, and replace the standard Mario sprites with dancing versions. I think it would be really cool to have a player’s gestures drive Mario’s movement, so that the embodied dance skills of the player become critical to leveling up in DMD!
Related Digital Arts MFA Work:
Mimesis & Mocap DANM-MFA Performance Study(Spring'10) (Thesis Committee: Kathy Foley, Michael Mateas, Ted Warburton) You can see a proof-of-concept of my PhD research direction by watching video documentation of my Digital Arts and New Media MFA thesis performances. I completed my MFA at UCSC in June 2010, consisting in part of a performance study (Mimesis & Mocap), that visualized the effect of making an autonomous agent mimic and improvise with an embodied actor through dramatic gesture. My live stage shows used pantomime and dance, within the comedic context of the classic Marx Brother's Mirror Gag, to create the illusion of expressive play between myself and my 3D Avatar. Here is a short preview video that shows excerpts from both The Avatar Dance group show and The Magic Mirror Game two-actor-one-man show:
|
|||
|
|
|||
|
|
|
|||
|
|
||||
|
|
||||
|
|
||||
|
|
||||
|
||||
|
||||
|
