> >This is another important issue.It's also harder to work around the problem >than you describe. I believe you'll need to perform the pick on a *set* of >single-element arrays, one for each element. This will more than double the >memory needed for the geometry, but worst of all can slow the performance of >the pick operation to the point that it becomes unusable when large enough >objects are involved. > >My suggestion to the J3D team is that the pick method should return three >additional integers in addition to the object hit. The integers are the >indices of: >1: the face or sub-element which was hit, >2: the nearest edge, and >3: the nearest vertex. > >It's possible that these concepts will not apply to all primitives in the >future. For example, a 3D text object wouldn't have any faces, edges or >vertices that the application programmer could identify, so for that object, >the returned "face" index should identify the character in the text string >that was hit, and no information for edge or vertex should be returned. For a >parametric patch object no face index would be returned, but the vertex index >can identify the nearest (or most influential) control point. IOW, these three >additional return values should map to the closest and most useful concepts to >those of "face", "edge" and "vertex". The documentation for each primitive >should describe its pick return indices. > >A purest might argue that this sort of concept overloading is a bit arbitrary >and that a much more complex system is needed for a 100% general solution, but >the one I suggest has the benefits of being extremely easy to implement and >completely solves the problem for at least 99% of applications. The fact that >this is not a perfect solution should not overly delay the inclusion of any >solution at all to this important problem.--Daniel Green >[EMAIL PROTECTED] >Superliminal Software 415-673-5965 >http://www.superliminal.com/ San Francisco, California > > > >From [EMAIL PROTECTED] Tue Apr 28 09:49:24 1998 >Return-Path: <[EMAIL PROTECTED]> >Received: by capra.eng.sun.com (SMI-8.6/SMI-SVR4) > id JAA10671; Tue, 28 Apr 1998 09:48:25 -0700 >Message-ID: <[EMAIL PROTECTED]> > I note that the last contribution on this topic was in April 98. I am urgently in need of a method of getting at least the approximate coordinates of the picked point on the object. Is there a Java3d method for getting this? Is there a similar discussion group elsewhere? Is there an update to the API about to be issued? Regards Mike =========================================================================== To unsubscribe, send email to [EMAIL PROTECTED] and include in the body of the message "signoff JAVA3D-INTEREST". For general help, send email to [EMAIL PROTECTED] and include in the body of the message "help".
