richard@gryphon.CTS.COM (Richard Sexton) (05/19/88)
In article <YWYRI-y00VAAA-Dlgc@andrew.cmu.edu> mp1u+@andrew.cmu.edu (Michael Portuesi) writes: >Richard Sexton expresses the following question: >> >> Huh ? > >Imagine you have a rectangle defined by its top/left and bottom/right >corner points: > > +------ > | | > ------+ > >You rotate an object by rotating its points: > > /\ > / + > / / > + / > \/ > Hmm. Okay. You are rotating all four points to generate the rotated rectangle. I still maintain that rectangles are most efficiantly represented with top left and bottom right points. It saves space in the display list and saves transmission time to a rendering host, if there is one. If you can compute the rotated points, shirley you can compute the "missing" vertices. If you cant, write me, I think I have the algorithm somewhere....... -- Have a nice day or Klortho will rip your nuts off. richard@gryphon.CTS.COM rutgers!marque!gryphon!richard
richard@gryphon.CTS.COM (Richard Sexton) (05/22/88)
(Ross Cunniff) writes: > >That would be an interesting algorithm indeed; I've always wondered if >anybody was working on psychic software :-). Seriously, two points do >not suffice if rectangles may be rotated. How can the software tell the >difference between these two rectangles: > > /\ > +----- + \ > | | \ \ (pretend the second one has all right angles) > -----+ \ + > \/ Oh. I thought you were rotating them after the fact, not rendering rotated rectangles. Never mind. I still say they're polygons, not rectangles :-) -- Have a nice day or Klortho will rip your nuts off. richard@gryphon.CTS.COM rutgers!marque!gryphon!richard