On Monday 21 March 2005 17:01, Timothy Miller wrote: > On Mon, 21 Mar 2005 13:33:48 +0100, Nicolai Haehnle <[EMAIL PROTECTED]> wrote: > > I vote for an explicit "continue trapezoid" command, or an equivalent > > mechanism that allows an arbitrary number of continuation trapezoids. The > > rationale is that this allows us to render some quads (and by extension, > > polygons) as three trapezoids instead of splitting them up into two > > triangles and ending up with four trapezoids. > > Daniel already mentioned the idea of doing r-to-l rendering for left > knee. And I already have to have it for proper bitblt. As it's > implemented, however, it's always the last pixel that's left off, so > for r2l, it would leave off the left pixel. If I tweak it so that you > can leave off the right pixel, even if it starts on the right, that'll > solve that problem for little cost. > > Any round-off issues here?
Don't think so. The situation is symmetrical after all. By the way, if you're going to do this, would it be easier to have one pixel pipe do the even x coordinates and the other the odd ones? Haven't thought it through, but I think it's been mentioned before for some reason. If you have to be able to turn both of them off anyway... Lourens _______________________________________________ Open-graphics mailing list [email protected] http://lists.duskglow.com/mailman/listinfo/open-graphics List service provided by Duskglow Consulting, LLC (www.duskglow.com)
