mbutts@RELAY.CS.NET (Mike Butts) (03/03/90)
>From article <1364@dgbt.uucp>, by andrew@dgbt.uucp (Andrew Patrick DGBT/DBR): > > In article <16386@well.sf.ca.us> 70117.2546@CompuServe.COM (Frank Dunn) writes: >>I'm wondering if we could start a discussion on applications of >>virtual-worlds technology in the telecommunications industry. > ... [ Stuff Deleted] > A broadcaster arranges to provide 'virtual world' coverage of a major > event -- say, the Canada Day celebrations on Parliament Hill in Ottawa > (or the fighting in Lebanon). First, the broadcaster would collect > 'stock' images and models of the physical layout, and some standard > interactions (e.g., imformation bases). The broadcaster might also > 'create' some virtual characters that can be used later, such as notable > historical figures. > > On the day of the event, the broadcaster would setup a number of live > cameras etc. to transmit the events as they happen. > > For the viewer (user) of this virtual-world broadcast, there would be > the capability of interacting with and moving about the live images, > as well as using the 'stock' information and historical characters. > *Yes* We may wish to think of this as "Object Oriented Television" (OOTV). > There are obvious questions about how that much information could be > transmitted during the event, but it is not impossible. The 'stock' > information could be transparently transmitted to the users and stored > before the event takes place. Thus, only the 'live' information need > be broadcast during the event, and there are already broadcasters > using multiple channels to cover the same event, with the user being > able to select amoung the different 'views'. Right. You transmit descriptions of the scene at the logical level, using that to position and orient locally-stored descriptions of the objects with respect to the local viewer. Much less bandwidth required. > What we are really lacking is the capability of 'integrating' > information that is stored locally with that coming in live. Small matter of compute power ;-). A much harder problem is doing the initial analysis of the live scene to generate the logical descriptions. At least for broadcasting, one could afford much hardware at the transmitting end. But how to do it? AI feature recognition of stereo video coupled with laser ranging? A gotcha might be: What happens when something totally unexpected (and unrepresented in the local object database) wanders into the scene? A sudden big spike in analysis and bandwidth may be required. > Also, we > are missing the human interface component that makes this interesting, > and easy to use. That's were the virtual-reality devices come into > play. Finally, at most importantly, we need to know what people want. > That is, we need to know what value systems like this will have for > people, and what they will be willing to pay for. People want to be where the action is. Mutually maximizing excitement and security is what most people want. Virtually being there via OOTV without having to travel, deal with the crowds, get shot at, pull the G's or whatever is very exciting. "You are there." It makes me yawn at the idea of HDTV. I'll bet we'll see OOTV in teleconferencing first. Simple scenes and corporate bucks. > So much to do... So little time... -- Michael Butts, Research Engineer KC7IT 503-626-1302(fax:1282) Mentor Graphics Corporation, 8500 SW Creekside Place, Beaverton, Oregon 97005 !{ogicse,sequent,tessi,apollo}!mntgfx!mbutts mbutts@pdx.MENTOR.COM Any opinions are my own, and aren't necessarily shared by Mentor Graphics Corp.