From drewk@graphics.cis.upenn.edu Thu May 20 22:15:51 1999 Received: from burdell.cc.gatech.edu (root@burdell.cc.gatech.edu [130.207.3.207]) by lennon.cc.gatech.edu (8.9.1/8.9.1) with ESMTP id WAA29429 for ; Thu, 20 May 1999 22:15:48 -0400 (EDT) Received: from asbestos.hitl.washington.edu (hitl-new.hitl.washington.edu [128.95.73.60]) by burdell.cc.gatech.edu (8.9.1/8.9.1) with ESMTP id WAA28159; Thu, 20 May 1999 22:15:45 -0400 (EDT) Received: from linc.cis.upenn.edu (LINC.CIS.UPENN.EDU [158.130.12.3]) by asbestos.hitl.washington.edu (8.9.3/8.9.3) with ESMTP id TAA20135 for <3d-ui@hitl.washington.edu>; Thu, 20 May 1999 19:15:20 -0700 (PDT) Received: from graphics.cis.upenn.edu (GRAPHICS.CIS.UPENN.EDU [158.130.2.10]) by linc.cis.upenn.edu (8.8.5/8.8.5) with ESMTP id WAA04535 for <3d-ui@hitl.washington.edu>; Thu, 20 May 1999 22:15:18 -0400 (EDT) Received: from roger.cis.upenn.edu (ROGER.CIS.UPENN.EDU [158.130.12.70]) by graphics.cis.upenn.edu (8.8.5/8.8.5) with ESMTP id WAA02675 for <3d-ui@hitl.washington.edu>; Thu, 20 May 1999 22:15:18 -0400 (EDT) Received: by roger.cis.upenn.edu id CAA03751; Fri, 21 May 1999 02:15:18 GMT Date: Fri, 21 May 1999 02:15:18 GMT Posted-Date: Fri, 21 May 1999 02:15:18 GMT Message-Id: <199905210215.CAA03751@roger.cis.upenn.edu> From: Drew Kessler MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit To: 3d-ui@hitl.washington.edu (3D UI List) Subject: Re: Virtual vs. real manipulation In-Reply-To: <199905181300.JAA21169@lennon.cc.gatech.edu> References: <199905181300.JAA21169@lennon.cc.gatech.edu> X-Mailer: VM 6.22 under 19.15 XEmacs Lucid Status: RO (Doug: good luck with your defense!) I think that this discussion about "realistic" vs. "magical" interaction in virtual environments is very interesting, informative, etc., but I feel that it really doesn't address the question posed to Doug, or, at least, my interpretation of the question. The way I read the question is: Given a specific task of placing a set of "grabable" objects (close and small enough) in some particular positional relation to each other, how much slower can a person perform this task in a virtual environment vs. performing the task in the real world? Yes, the motivation for doing the task is an issue, but the question asks about speed and accuracy, not effectiveness in training on a task. My read on this is that the questioner would allow the VE app to use the most effective interaction technique, be it "realistic" or "magical". What I think is interesting is that the questioner assumes that the task can't be performed as well, or faster in a VE than in the real world. This is clearly not true: If the task is to place three or more objects in specific, but arbitrary, unsupported spots in 3-space, the VE wins hands down (no pun intended :) To get a more meaningful answer, I think that the task needs to be more specific. For example: given a set of real and virtual Lego blocks, how fast can someone build a model that they have built many times before? The model is arbitrary, the set of bricks is arbitrary. The "magical" command, "build the model" isn't available, because the model isn't known by the app. Commands such as "place the blue brick on the red brick" are likely to be too ambiguous. I would imagine that some combination of magical and direct interaction techniques would provide the best performance. Better than the real world? Maybe not, but I remember seeing a Lego design application being used by an expert at the Digital Biyou, SIGGRAPH '96, where models were being built very quickly. (Biggest gain would be in the "search for piece" task, I'm guessing.) I do agree with Matt's bottom line claim: "Bottom line claim I make: you can ask how close any *given* Realistic VR experience is to its analogous Real Life Experience, but I don't see how you can make broad claims about how the medium of VR in general is or isn't close to replicating real life experiences in general." But, I think you can identify specific tasks that can be performed, perhaps using different methods, by both VE apps and using tools and props in the real world, and make a meaningful comparison that gets to what I believe is the root of the question at hand: "Why did I spend all this money on VE stuff, again?" Ok, so I have no experiments to back this up, but I think I can answer Doug's questions as follows: > In terms of performance (speed and errors), how far are we > from real-world performance on a simple, within-reach object > manipulation task in VR? That is, if you have a set of 4 blocks > that you want to place one on top of the other, how much worse > is it in VR than in real life? Real-world performance will generally be faster, except when objects are placed with no support or need to be glued to stay where they are, or when extremely high accuracy is required (within 0.01 mm?). > Are we close to the optimal we can do in VR? No, of course not, but we aren't optimal in the real world either. Obviously, a device could be built to assist people in placing objects more accurately, or floating in space. > If not, what would cause performance to leap to the next level? Identification of specific task, design of interface appropriate to the task (with the right "magical" commands). > Are these even the right questions to be asking? Maybe (see previous discussion)... Take care, all, -Drew -- ___________________________________________________________________________ G. Drew Kessler drewk@cis.upenn.edu Dept. of Computer and Information Science Office: 174 Moore Bldg University of Pennsylvania Phone: (215)573-2815 200 South 33rd Street Fax: (215)898-0587 Philadelphia, PA, USA 19104-6389 http://www.cis.upenn.edu/~drewk/home.html