From: "Bruce Cohen;;50-662;LP=A;" Subject: Re: VR in art - medium or instrument? (sculpture) Date: 30 Nov 90 18:05:49 GMT Organization: Tektronix Inc. In article <12012@milton.u.washington.edu> mg@munnari.oz.au (Mike Gigante) writes: > This of course is the big question. I have though of surrogate methods > for feedback. One of them is a superposition of a regular grid in the > working space. Anytime part of the object or hand passes through the > any of the cell walls, a projection of the wall is superimposed on the > object/hand. (Is this clear? it is sort of like passing your hand through > a `force-field' as in sf movies.) If these cell boundaries are of > different colours, you can at least tell proximity. It doesn't help as > much as physical feedback, but I don't know much about that area. Your description's clear enough. That's one of the techniques we came up with here at Tek when we were trying to develop better visual feedback for the monocular 3D display we sold on our graphic workstation. The problem there was trying to see where the cursor is in 3 space as you drive it around with your 6 DOF input device. With stereo it's not too hard, but monocular requires additional depth cues and/or intersection cues. The wall intersection idea works OK as long as the visual aspect of the wall's projection on the hand doesn't interfere with the object being manipulated. So the wall's color can't be too opaque (or it has to be a relatively coarse mesh), and it's texture can't beat destructively with the object's texture. The difficulty with any of those techniques is, of course, that they're visual, and sculpting is a kinesthetic activity (if you doubt that, give almost anyone a blob of clay and see what she does with it; most people roll it out their hands like sausage for the tactile sensation of it). > > One of the other possibilities is little nodules that can be inflated/raised > (or whatever) when you `touch' some VR object. I saw a mouse at siggraph > that had something like that. > I've been playing around with a lot of ideas for tactile feedback; there don't seem to be any technologies which are really well-suited to the job, either because they require lots of apparatus surrounding the user, like the NASA hand, which would probably put a lot of people off, or because they require manufacturing techniques which are currently only one-off, like making a glove with several hundred rapidly inflatable/deflatable pockets on the fingertips and palm. As far as I can see there are three different types of feedback, with different uses and requirements: 1) Position feedback - used to prevent movement through a volume in which there is a virtual object. You could consider it an extreme case of type 2, but so far all the implementations I've been able to dream up require seperate mechanisms. One way to fake this is to have real objects taking up the same space as the virtual ones, and just use the VR interface to change the aspect of the objects (like the phoney Ming vase I mentioned in a previous posting). It's a limited fake, though. 2) Force feedback - used to feed back resistance to movement, either from viscosity in the medium (e.g., moving around underwater), resistance of an object to being moved from inertia or friction, or resistance of a material to being deformed (this is the primary use in sculpting). This category is actually two areas: global, which involves the whole body's interaction with the medium, and local, which could just involve the hands and the virtual objects they're holding. 3) Tactile feedback - used to provide small-scale information on surface texture. This is the really nasty one, because it requires lots of effectors. Unfortunately for this discussion, it's also useful in sculpting, because you do want to be able to sense and control the finish of an object. High quality rendering with specular reflection and good texture reproduction would help in seeing the finish. > UNC's system of active force feedback doesn't seem quite so relevent in > the sculpting case. I dunno, maybe we need an active `straight-jacket' that > you wear. using inverse kinematics, you could constrain the hand position. > yet still allow elbow movement etc. In fact I like this idea...> Yep, that's the way to solve the global problem for force, *if* people are willing to put on a straight-jacket. To handle position, you still have to anchor the straight-jacket in absolute coordinates. -- ------------------------------------------------------------------------ Speaker-to-managers, aka Bruce Cohen, Computer Research Lab email: brucec@tekchips.labs.tek.com Tektronix Laboratories, Tektronix, Inc. phone: (503)627-5241 M/S 50-662, P.O. Box 500, Beaverton, OR 97077