From: owner-3dui@hitl.washington.edu on behalf of Robert W. Lindeman [gogo@seas.gwu.edu] Sent: Thursday, October 04, 2001 3:12 PM To: 3D UI list Subject: Re: text/number input in VEs Hi all, First, I have to congratualet Doug for spawning what is probably the longest thread to date on the 3DUI List! So many messages, that I was astounded when I arrived and logged in today! I also find this to be an interesting topic. We use HMD environments as well. If we take a more-global view of providing an infrastructure for VE interaction (a la windowing environments in 2D), then textual/numeric input is certainly atractive. *That* it is important for some applications, I think we can agree upon. *How* this input is made is the problematic point. Just as in WIMP environments there are multiple ways of carrying out a give action, we should have multiple ways of doing things in VEs. For example, one can copy a file by 1) drag-and-drop, 2) cut-and-paste, 3) entering a voice command, 4) entering a command on the command line, or 5) using eye-tracking and dwelling. Different *users* will use different methods, and the same user will use different methods at different times, or for different tasks. This is an old story, but the user-task-environment tension must be taken into account. We published work on using 2D interfaces in 3D in IEEE VR'99 & CHI '99. In our testbed, we also implemented a keyboard interface (mainly just for fun), by displaying a QWERTY keyboard on an interaction paddle. Those of you with a copy of the video proceedings of these two conferences, I think we demo the keyboard interface in our video submission. I think the guys in Dieter Schmalstieg's group at TU-Wien are doing similar stuff (Personal Interaction Panel, or PIP) for rear-projection environments. In short, you hold a paddle in one hand, superimpose widgets (in this case keyboard buttons) on the paddle in the visual field, and select them using a stylus or your finger. This is not very fast, but for short annotations, numeric input, and the like it might be okay. Also, because the paddle in moveable, it is there when you need it, and gone when you don't. It can support different widgets on the same panel, so it can display context-specific widgets. Finally, it can be held in the same field of view as the object(s) you are modifying, so a toolglass approach is also possible. I like the idea of supporting multiple widget types in a unified framework. It might make life easier for us all if there was some sort of common API for coding this stuff up. Anyway, that is my US$0.02. Flame on! -Rob --- Dr. Robert W. Lindeman The George Washington University Assistant Professor Department of Computer Science E-Mail: gogo@seas.gwu.edu Web: http://www.seas.gwu.edu/~gogo/ "Chi non risica, non-rosica."