Re: Soft for Dancers Up Date #2/ Articles On Line[wasDear Richard, I try in a few words to sketch a my (partial) answer on "why" and "which outputs" we aim to achieve at our lab. - research: we aim at understanding which cues and which invariants better identify movement expressive and emotional qualities as perceived by spectators. Our proposed idea of working on "Microdances" is the first step of our approach: a lab testbed for analysis of data on real movement fragments (not necessarily only in terms of MoCap data - since a spectator does not perceive these data as input, to say very roughly). We also work on a hypothesis different from MoCap: not a controlled (light etc) environment with markers, but a natural setting or a theatre stage in real-time. (This choice comes also because we need in our microdances to have a "natural" setting where an artist feels as less perturbated as possible in order not to influence/disturb his/her expression). This microdance testbed is used to experiment and test new algorithms and techniques for extraction of relevant cues from movement, and to evaluate/validate the outputs from such techniques: I mean that if results from algorithms are coherent with data from spectators' rating on the same microdances (i.e. using the same stimuli as inputs, i.e. the same microdances) then this can be a concrete validation method of methods of analysis of expressiveness. The outputs from research are therefore: which cues are relevant w.r.t. certain expressiveness/emotional in certain context, and which "shapes" and behaviors such cues can help to measure. Which invariants, etc. I might say more but I do not want to take too much space in the list. Please ask me directly or please have a look at our paper in our web site infomus.dist.unige.it (select "Publications" page, the first PDF linked to the list, of 2002) in case you wish more info. Very shortly: Another example of challenging research direction: similar cues can be considered also in the analysis of audio (music) signal: expressive cues as a means of better understanding cross/multi-modal mappings, with benefit for HCI in general? - artistic output: what we learn on microdances and in long Lab sessions with artists is then engineered on libraries for our EyesWeb platform. Our aim isto contribute to overcome current limitations of state of the art "interactive systems": e.g., new models of interaction (e.g. beyond the "musical instrument" metaphor), new models of (cross-modal, multimodal) "mapping", etc. We developed the EyesWeb open software platform not only to support our (and others') research, but also to experiment in these directions with artists in real artistic productions. We included in EyesWeb some support to easier implementation of such scenarios (eg active modules mechanism, real-time scheduler etc). We distribute for free the full version of EyesWeb (not a prototype, a real engineered robust sw) in order to share our developments and possibly to learn/exchange with others in these directions. - potential commercial outputs of these research directions (and of Eyesweb developments) include new/mixed-media, broadcasting, multimedia apps in general. In our case, we find resources for our research both from research contracts and from contracts with industry and institutions (theatres, museums, science centers, artists, etc) to develop concrete interactive installations, support artistic productions, etc. (see our web page) based on EyesWeb. All the best Antonio ----- Original Message ----- From: Richard Widgery To: dance-tech@lists.acs.ohio-state.edu Sent: Wednesday, April 10, 2002 11:59 PM Subject: RE: Soft for Dancers Up Date #2/ Articles On Line[was dance and language thread] Dear Armando & Antonio, May I thank you for your narratives; this discussion is really getting interesting, sorry I have not joined in before... partly I wished to see how it would pan out before jumping in. It is now getting very wide based, multi threaded and wish to address a "few" issues. I tend to find more agreement with Armando's replies in respect to "Movement vs. Gesture", "Velocity Peeking", "Emotional Engagement", and "Psychological Archetypes", but now wish to diverse into what I consider to be a more fundamental principal of software development: "What are you trying to achieve?". "Given enough resources - I can develop anything...." seems to be a common drive of mankind. I have just had the 'timing' fortune to see Spielberg's (aka Kubrick's) A.I. feature film last night, which unintentionally (i.e. not script writer intention) brought up some very serious questions about commercial exploitation that they seemed to overlook... mainly WHY create a sentient avatar - regardless of Hollywood happy ending stories. I had a discussion with a well known Visual Effects company a few years back when they asked me how long it would be until I could generate a fully synthetic, visually unquestionable, representations of a human actor. When I asked why, they replied because the client did not want to pay background extras (actors) royalty right every time they aired the commercials. OK.. I'm not going to go off one a commercial exploits argument, but what I would like to discuss is the INTENTIONS / APPLICATIONS of the types of software you would like to use or create. Everyone so far has put forward very good and constructive aspects of tools development without ever saying HOW they want to use them (Armando / mjumbe describe methods of implementation via scripting, but never describes the application). As a software developer, it is essential to know know why people are using a tools as much as knowing how, and I would like to ask if you would ideally like a computer to totally synthesis it's own dance choreography, or whether you would like better tools for pre/visualise 3D animations of your own performances, or whether they would like to have computer augment visual/audio performances in Real/Non Real time...etc... What are the fundamental OUTPUTs you are trying to achieve? I agree that a library of "microdances" is 'almost' essential for any of the above, but what interest me more as a MoCap expert is why you would go to the trouble of recording 50 microdances, bolting them together (with some emotional variables/blending factors) and generating a 3D representation of a continuous dance segment of performance - rather than just capturing the thing in the first place. This comes back to the discussion about emotional content, and I propose that a singular performer (human real person) will give a more congruent, sensitive, comprehensive, and component compliant performance with "tonal" harmony than a piece-meal compilation of microdances. If you disagree - I would ask WHY would you WANT a computer to produce a congruent.........harmony performance all by itself. Are you doing it because it is there to be done? Or are you doing it for personal artistic merits? If you fall into the first category I would like to ask where you think the boundaries of A.I. movement interaction should become a moral debate (please do not reply that Art & Morals should not be discussed together - unconstructive), and if you fall into the latter - could you please be more specific about you "product" or "presentation" that you wish to achieve. I feel that what I have written my fallen short of expectations of "artists", but I whish to get a better grasp of the OBJECTIVE and QUALITATIVE expectations of digital choreographers. For example I was pleased to see Merce Cunningham's "Biped" presentation at the Barbican - London last year - I was impressed with much of the graphical interpretation of body movement to abstract illustration / graphic art, but disappointed when a "raw video representation" of mocap markers came flooding across the screen - for me this is like seeing the dust on the inside of a cathode ray tube and someone stating "wow - look at the art in the fact we are using optical motion capture markers (i.e. ping pong balls)" - sorry it a part of the process I see every day, day in, day out, not piece of art just because it has been put into a context of an auditorium (personal critique - sorry).9 In terms of software, visualisation / presentation, tools development, what OUTPUT / RESULT would you like to achieve - maybe debate this before debating what process / language you would implement? Food for thought, Richard Widgery richardw@kinetic-impulse.com -------------------------------------------------------- KINETIC IMPULSE Digital Performance Art & Science http://www.kinetic-impulse.com --------------------------------------------------------
This archive was generated by hypermail 2b30 : 04/13/02