From: owner-3dui@hitl.washington.edu on behalf of Raimund Dachselt [dachselt@inf.tu-dresden.de] Sent: Friday, August 11, 2000 11:34 AM To: 3d-ui@hitl.washington.edu Subject: Re: Topic: dynamically creating (copies of) objects Hi Jeff (et al.), I enjoyed reading your thoughts on the topic of object copies in virtual worlds. With your detailed ideas you really did not leave much room for us to throw in new ones... Concerning the interaction techniques I once developed a system, where a technique similar to number 7 was used. In a desktop VR system a car could be manipulated in order to change details or find specific parts to get information about them / to order them. Different copies of objects for different manipulations / operations were successfully used in a single user application. It seems to be useful to assign certain operations to particular copies of the same object, where the copies don't have to be exactly identical but can differ in size and other properties to support the associated tasks well. In more detail: A car was displayed in the center of the screen with which you could interact in various ways (change colors, select single parts, change transparencies and so on). A small (and - due to performance reasons - simplified) smaller copy of the car in the corner of the screen was only used for the rotation of the original car and to set special views (front, top, side etc.), thus working as a separated trackball. Both objects are synchronized in their rotation, so manipulating the rotation of the small one changed the big, whereas operations like changing colors were only displayed for the original. Thus the small car was used as a visual "object reference". There are several advantages of this technique: - the reduction of the number of operations on the (original) object through the clear separation of viewpoint manipulation and object interactions - no obscuration of the original object (which would occur with trackballs or similar transformation widgets) - besides free rotation additional pre-defined viewpoints can be set in an intuitive way with the help of the mini-model (that means just simply clicking on the side of the car gives you a side view and so forth) - Probably the most important: Since the original car could heavily be changed in appearance (parts be removed or made transparant so that only say the braking system was visible) it would have been difficult to still know the position of the remaining parts within the car. The copy gives you a clear sense of the parts' position. By the way, allowing to split operations in main operations on the big object and secondary operations on the smaller copy / copies results from the constraint of one-handed interaction. If you had say a spaceball in your non-dominant hand, you could use it for rotation of the object and its buttons to set standard views, whereas the dominant hand would manipulate the object, usually with the mouse. Even in this two-handed scenario it visually helps to see the small copy. Sorry for the long text. Have a good weekend, Raimund __________________________________________________________ Raimund Dachselt Dresden University of Technology - Department of Computer Science Endowed Chair for Multimedia Technology D-01062 Dresden Phone: +49 351 463-8514 Fax: +49 351 463-8518 E-Mail: dachselt@inf.tu-dresden.de http://www-mmt.inf.tu-dresden.de/team/Person.asp?id=rd8